Latest / Connor with Honor: AI, Real Estate and Real Life / Only 9% of Americans Are Excited About AI. I Built a Case File Anyway.

Transcript

Only 9% of Americans Are Excited About AI. I Built a Case File Anyway. · Connor with Honor: AI, Real Estate and Real Life

• 0:00
  SPEAKER_00: Well, my name's Connor McIver.
• 0:02
  SPEAKER_00: I spent 23 years carrying a badge in Los Angeles.
• 0:05
  SPEAKER_00: I've sold real estate since 1998.
• 0:07
  SPEAKER_00: I'm 57 years old.
• 0:08
  SPEAKER_00: I'm a Gen Xer, of course.
• 0:10
  SPEAKER_00: I still write checks every month.
• 0:12
  SPEAKER_00: Well, I would write a check if I had the ability to get health insurance, but I still flinch uh thinking about the perspective and ability of maybe getting it one day.
• 0:22
  SPEAKER_00: But I have to plan to be here for another 500 years.
• 0:26
  SPEAKER_00: That's my desire.
• 0:27
  SPEAKER_00: So now before you close this, let me tell you what this is.
• 0:30
  SPEAKER_00: So over the last several weeks, I pulled 300 videos from YouTube.
• 0:34
  SPEAKER_00: Every futurist I could find talking about artificial intelligence and where the human race is headed.
• 0:40
  SPEAKER_00: 300.
• 0:41
  SPEAKER_00: I watched, I sorted, I took notes.
• 0:43
  SPEAKER_00: Now, might be surprised.
• 0:44
  SPEAKER_00: There's a lot of people out there that think it's a bad thing, think it's going to end the world.
• 0:48
  SPEAKER_00: But I'd rather walk into this thing positive, thinking that it's all going to work out in some way, shape, or form.
• 0:54
  SPEAKER_00: And if I start to see it's not, well, that cop in me says need to take immediate action.
• 0:58
  SPEAKER_00: But right now, I don't know if what either side is telling us is true.
• 1:03
  SPEAKER_00: But I thought I would kind of play a futurist for a moment and think, well, what if this is the best thing that's ever happened to the human race?
• 1:10
  SPEAKER_00: And then if that's the case, what does that look like?
• 1:13
  SPEAKER_00: Kind of pull all the experts into one space and see what it is.
• 1:17
  SPEAKER_00: I went looking for the good news, couldn't find it anywhere else.
• 1:20
  SPEAKER_00: So here's my bias stated up front so you can weigh everything, and this will be a little bit repetitive, but please appreciate it.
• 1:27
  SPEAKER_00: I would rather be positive facing certain tragedy than negative headed toward abundance.
• 1:32
  SPEAKER_00: That's the whole sentence.
• 1:33
  SPEAKER_00: That's the whole show.
• 1:34
  SPEAKER_00: If I'm wrong, I will have spent my last years very hopeful.
• 1:37
  SPEAKER_00: I think that's a better way to go out of any existence.
• 1:40
  SPEAKER_00: If the doomers are wrong, well, they will have spent theirs afraid of nothing.
• 1:45
  SPEAKER_00: And I know which of those two mistakes I'd rather make, as stated.
• 1:49
  SPEAKER_00: But I'm not going to ask you to feel your way to that.
• 1:52
  SPEAKER_00: Of course, I'm a cop.
• 1:53
  SPEAKER_00: I built cases for a living, wrote a lot of tickets, and that's what this is.
• 1:58
  SPEAKER_00: Not a pep talk, it's a case file, 30 pieces of evidence, every one of them with citation, go check yourself, and cross-examination at the end, where I put my own argument on the stand and I try to break it.
• 2:08
  SPEAKER_00: So now we're going to start with a number.
• 2:11
  SPEAKER_00: And that's going to be 9%.
• 2:16
  SPEAKER_00: And this is 9% of American adults say that they're more excited than concerned about artificial intelligence in their daily life.
• 2:25
  SPEAKER_00: 52% say they're more concerned than excited.
• 2:29
  SPEAKER_00: That's a big difference.
• 2:30
  SPEAKER_00: Now, this is Pew Research Center, 3,488.
• 2:34
  SPEAKER_00: American adults surveyed in June of this year.
• 2:37
  SPEAKER_00: Not a vibe, not a hot take, but a property-weighted national sample from the most careful pollster in the country.
• 2:45
  SPEAKER_00: 6-1.
• 2:46
  SPEAKER_00: Fear to hope, that's 6-1.
• 2:48
  SPEAKER_00: Now I want you to sit with something.
• 2:49
  SPEAKER_00: When I started this project before I read a single study, I guess the split in those 300 videos was about 10% good news and about 90% bad.
• 2:59
  SPEAKER_00: That was my gut.
• 3:00
  SPEAKER_00: That was a cop's gut ever watching after watching a few hundred people talk about the future.
• 3:05
  SPEAKER_00: Pew says nine.
• 3:06
  SPEAKER_00: My gut was off by a point, and that's not because I'm clairvoyant.
• 3:09
  SPEAKER_00: It's just because when a thing is lopsided, you can feel it without measuring it.
• 3:14
  SPEAKER_00: You walk into a room, you know something's wrong before anybody tells you, just by the feeling in the air without really any other reason, especially if you played cop for any amount of time.
• 3:23
  SPEAKER_00: Now that's the last four years of AI coverage.
• 3:26
  SPEAKER_00: That's what it's felt like to me.
• 3:28
  SPEAKER_00: So this show is for the 9%, and more importantly, it's for the 37% in the middle who are actually on the fence and only ever hear one side of the argument.
• 3:40
  SPEAKER_00: So the single number is not the story, it's actually the trend that's the story.
• 3:44
  SPEAKER_00: So look at these two lines.
• 3:45
  SPEAKER_00: So in 2021, 37% of Americans were more concerned than excited.
• 3:50
  SPEAKER_00: Today it's 52.
• 3:52
  SPEAKER_00: Concern went up 15 points.
• 3:54
  SPEAKER_00: And that's bad, but it's not shocking.
• 3:56
  SPEAKER_00: New technology, people get nervous.
• 3:58
  SPEAKER_00: We've all seen that movie.
• 4:00
  SPEAKER_00: Now, look at the other line, the hopeful line.
• 4:02
  SPEAKER_00: In 2021, 18% of Americans were more excited than concerned.
• 4:07
  SPEAKER_00: Today that's nine.
• 4:09
  SPEAKER_00: 2022 November was when ChatGPT 3.5 made its debut.
• 4:14
  SPEAKER_00: That was the end of November.
• 4:16
  SPEAKER_00: And uh it's been moving real quick ever since.
• 4:20
  SPEAKER_00: In fact, lately, artificial general intelligence, yeah, that's what they're saying.
• 4:25
  SPEAKER_00: Now, hope didn't slide.
• 4:27
  SPEAKER_00: Hope got cut in half.
• 4:28
  SPEAKER_00: And here's the part that stopped me cold.
• 4:30
  SPEAKER_00: That collapse happened during the exact five years and when most people in history started actually using these tools.
• 4:36
  SPEAKER_00: Usage went completely vertical.
• 4:38
  SPEAKER_00: Enthusiasm got cut in half.
• 4:40
  SPEAKER_00: People are using AI more and liking it less.
• 4:44
  SPEAKER_00: And it's worst amongst the young.
• 4:47
  SPEAKER_00: For the first time on record, a majority of American adults under 30, 55% are more concerned than excited, up from 31% in 2021.
• 4:55
  SPEAKER_00: The generation using it the most is the generation most afraid of it.
• 4:59
  SPEAKER_00: That's not a technology problem.
• 5:01
  SPEAKER_00: The technology got better every one of those years.
• 5:04
  SPEAKER_00: It's a story problem.
• 5:05
  SPEAKER_00: Somebody has been telling them a story.
• 5:08
  SPEAKER_00: Now let me show you who's paying for that story.
• 5:12
  SPEAKER_00: Now I have to correct myself here on camera, the first 10 minutes, and I want you to watch me do it.
• 5:20
  SPEAKER_00: My hypothesis going in was about 10% of these videos would be hopeful and 90% would be doomed.
• 5:26
  SPEAKER_00: That was my gut.
• 5:27
  SPEAKER_00: I said it out loud to people before I even checked.
• 5:30
  SPEAKER_00: So I checked.
• 5:31
  SPEAKER_00: I had every source classified by stance, optimistic, pessimistic, or neutral and mixed.
• 5:36
  SPEAKER_00: 257 of them got clean classification.
• 5:39
  SPEAKER_00: But here's what came back.
• 5:40
  SPEAKER_00: So optimistic, you got 21 sources, that's 8.2%.
• 5:45
  SPEAKER_00: Pessimistic and fear-based, 23 sources, 8.9%.
• 5:49
  SPEAKER_00: Neutral or mix, 213, 82.9%.
• 5:53
  SPEAKER_00: So I was wrong.
• 5:54
  SPEAKER_00: Doom and hope were almost dead even.
• 5:57
  SPEAKER_00: And you know what?
• 5:59
  SPEAKER_00: It's always always social media being fed by AI got me right there.
• 6:03
  SPEAKER_00: I was wrong.
• 6:04
  SPEAKER_00: Doom and hope were almost dead even.
• 6:07
  SPEAKER_00: It's crazy.
• 6:08
  SPEAKER_00: 23 to 21.
• 6:10
  SPEAKER_00: And the overwhelming majority of what people are actually publishing about this is neither.
• 6:15
  SPEAKER_00: It's technical, it's balanced.
• 6:18
  SPEAKER_00: It's somebody reviewing a model or arguing a thought experiment without picking a sign.
• 6:24
  SPEAKER_00: So the material is not the problem, which means the imbalance is happening somewhere between the material and your melon, your head.
• 6:31
  SPEAKER_00: And that's much more an interesting finding than the one I actually went looking for, and kind of makes me a little happy.
• 6:39
  SPEAKER_00: So here's the gap in one line.
• 6:40
  SPEAKER_00: The videos are roughly balanced.
• 6:42
  SPEAKER_00: The audience is six to one afraid.
• 6:44
  SPEAKER_00: Something in between is doing that.
• 6:46
  SPEAKER_00: Let me show you exactly what, and I actually have the receipt.
• 6:50
  SPEAKER_00: So this is the most important number in the show, and almost nobody's heard of it.
• 6:54
  SPEAKER_00: Now, researchers got their hands on an Upworthy archive, Upworthy Rant, headline A B test at enormous scale.
• 7:02
  SPEAKER_00: 22,743 randomized controlled trials, over 100,000 headline variations, 370 million impressions.
• 7:13
  SPEAKER_00: Here's what they found published in Nature Human Behavior in 2023.
• 7:17
  SPEAKER_00: Every single negative word you add to a headline raises the click rate by 2.3%.
• 7:24
  SPEAKER_00: Every single positive word you add lowers it by 1%.
• 7:29
  SPEAKER_00: Now read that again.
• 7:30
  SPEAKER_00: Adding a positive word to a headline actively costs you readers.
• 7:35
  SPEAKER_00: And I want to be precise about why this matters more than every media criticism you've ever heard.
• 7:40
  SPEAKER_00: It's not a survey, and this isn't somebody's content analysis where they decided what counted as a negative.
• 7:47
  SPEAKER_00: This is 22,000 actual randomized experiments.
• 7:51
  SPEAKER_00: That's casual.
• 7:52
  SPEAKER_00: That's the same standard we used to approve drugs.
• 7:55
  SPEAKER_00: So now the beautiful part.
• 7:57
  SPEAKER_00: Upworthy was the positive positivity website.
• 8:01
  SPEAKER_00: Their entire brand was uplifting news.
• 8:03
  SPEAKER_00: And even there, in the lab of optimism, one machine paid better for fear.
• 8:09
  SPEAKER_00: Now there's a second study in Scientific Reports, 95,000 articles measured up against 579 million social media posts.
• 8:18
  SPEAKER_00: Negative articles, 579 million social negative articles were 1.91 times more likely to be shared.
• 8:27
  SPEAKER_00: On Facebook, 98% more posts.
• 8:30
  SPEAKER_00: So let me say that quiet part out loud.
• 8:32
  SPEAKER_00: Nobody has sat in a room and decided to scare you about AI.
• 8:36
  SPEAKER_00: It's actually worse than that.
• 8:37
  SPEAKER_00: Nobody had to.
• 8:38
  SPEAKER_00: The incentive did it automatically at scale one A B test at a time.
• 8:44
  SPEAKER_00: So here's what that incentive produced.
• 8:46
  SPEAKER_00: Two researchers, Roe and Perkins, hand-coded 671 artificial intelligence headlines across seven British news national newspapers.
• 8:57
  SPEAKER_00: Hand-coded, not a sentiment algorithm, human beings doing the reading and the sorting.
• 9:03
  SPEAKER_00: The single largest category, 37%, was impending danger, 248 headlines out of 671 framed around the threat.
• 9:13
  SPEAKER_00: The smallest category, 11%, was positive capability.
• 9:17
  SPEAKER_00: 3.41.
• 9:19
  SPEAKER_00: Well, this is where going, I'm going to do something that the Doom Crowd never does for me.
• 9:25
  SPEAKER_00: I'm going to give you the evidence against my own point.
• 9:28
  SPEAKER_00: There are two other studies out there using automated sentiment tools on the full text of AI articles, and they find AI coverage is mostly neutral and slightly positive.
• 9:37
  SPEAKER_00: So if I hid those from you, somebody would find them and use them to throw out everything else I'd say.
• 9:42
  SPEAKER_00: Or fear does not dominate the total volume of AI writing.
• 9:45
  SPEAKER_00: Fear dominates the headline.
• 9:46
  SPEAKER_00: Fear dominates the share.
• 9:48
  SPEAKER_00: Fear dominates the thumbnail.
• 9:49
  SPEAKER_00: Fear dominates what makes it into your head, which is the only part that matters, really, because almost nobody reads past the headline.
• 9:57
  SPEAKER_00: All right, so that's the diagnosis.
• 9:59
  SPEAKER_00: Now let me open up the case file.
• 10:00
  SPEAKER_00: I'm going to start with things that have already happened.
• 10:02
  SPEAKER_00: Past tense, not a promise, not a projection, but done.
• 10:06
  SPEAKER_00: Everything in the next 15 minutes has already happened.
• 10:10
  SPEAKER_00: I'm not going to tell you what AI might do.
• 10:12
  SPEAKER_00: I'm going to tell you what it did with a date in a journal and a patient count.
• 10:17
  SPEAKER_00: If any of it's wrong, you can email me and I'll correct it on camera.
• 10:20
  SPEAKER_00: I sorted these the way I'd sort a case.
• 10:22
  SPEAKER_00: Strongest evidence first.
• 10:24
  SPEAKER_00: Strongest evidence in the entire file isn't about money or productivity or the stock market.
• 10:29
  SPEAKER_00: It's about a man who got his voice back.
• 10:32
  SPEAKER_00: Casey Harrell.
• 10:34
  SPEAKER_00: So 45 years old ALS patient.
• 10:37
  SPEAKER_00: That's the disease that took Luke Gehrig and Steve Stephen Hawking.
• 10:42
  SPEAKER_00: By 2023, his speech was gone.
• 10:44
  SPEAKER_00: His wife, his caregiver, and his caregiver could sometimes make out what he was trying to say, and that was sometimes.
• 10:51
  SPEAKER_00: Researchers at UC Davis put electrodes in his brain and trained a model on his neural signals.
• 10:57
  SPEAKER_00: 30 minutes of training, 99.6% accuracy on a 50-word vocabulary in 30 minutes.
• 11:05
  SPEAKER_00: Then 90 more minutes of training, 90.2% accuracy on 125,000-word vocabulary, and that's the whole English language.
• 11:14
  SPEAKER_00: Sustained accuracy after continued use, 97.5%.
• 11:19
  SPEAKER_00: New England Journal of Medicine, August 14th, 2024.
• 11:22
  SPEAKER_00: That's the top medical journal on the planet.
• 11:25
  SPEAKER_00: I want to tell you why this one hit me.
• 11:26
  SPEAKER_00: In 23 years of policing, the calls that never left me were not the violent ones.
• 11:32
  SPEAKER_00: They're the ones where somebody needed to tell me something and they couldn't.
• 11:35
  SPEAKER_00: The stroke victim, a kid too scared to speak.
• 11:38
  SPEAKER_00: The old man on the floor of his kitchen looking at me, trying to form a word.
• 11:44
  SPEAKER_00: It's a specific kind of horror watching a person locked inside of themselves.
• 11:48
  SPEAKER_00: A machine opened that door in 30 minutes, and I have to sit here and be told AI is a threat to humanity.
• 11:55
  SPEAKER_00: Maybe so.
• 11:56
  SPEAKER_00: It didn't stop there.
• 11:58
  SPEAKER_00: Exhibit number seven.
• 12:02
  SPEAKER_00: A woman named Anne, paralyzed by a brainstem stroke in 2023.
• 12:07
  SPEAKER_00: A team at UCSF decoded her speech at 78 words per minute and synthesized it in her own voice, rebuilt from a recording of her wedding video from before the stroke.
• 12:18
  SPEAKER_00: Now her voice, the one that her husband actually married, but there's a lag about eight seconds between her thinking, a sentence, and a sound coming out.
• 12:27
  SPEAKER_00: You can't have a conversation with an eight-second delay.
• 12:30
  SPEAKER_00: You can issue statements, but you can't really talk fluidly.
• 12:34
  SPEAKER_00: April of 2025, same participant, same team, they got it under one second streaming.
• 12:40
  SPEAKER_00: Nature Neuroscience.
• 12:42
  SPEAKER_00: And she went from broadcasting to talking.
• 12:45
  SPEAKER_00: And then August of last year, Stanford and Emery decoded in her speech, not attempted speech, but imagined speech.
• 12:53
  SPEAKER_00: Words the participants never tried to say out loud, up to 74% accuracy on 125,000 word vocabulary.
• 13:00
  SPEAKER_00: And here's the detail that tells you these are serious people.
• 13:03
  SPEAKER_00: They built a mental password.
• 13:05
  SPEAKER_00: The participants had to think a specific phrase, chitty chitty bang bang, to unlock the decoder.
• 13:12
  SPEAKER_00: It worked over 98% with 98% reliability.
• 13:16
  SPEAKER_00: They solved mind reading, and their first move was to install a lock on it.
• 13:21
  SPEAKER_00: The Doom Story, nobody's thinking about safety.
• 13:24
  SPEAKER_00: The people actually building this stuff put a password on a thought.
• 13:29
  SPEAKER_00: Exhibit 8.
• 13:30
  SPEAKER_00: This is the one I would lead with in front of a jury.
• 13:34
  SPEAKER_00: A protein is a chain of amino acids that folds itself into a shape.
• 13:38
  SPEAKER_00: These 3D shapes, they're everything.
• 13:40
  SPEAKER_00: It's a shape that decides whether it's going to fight our virus or it's going to cause Alzheimer's.
• 13:45
  SPEAKER_00: Figuring out that shape used to take a PhD student years, sometimes an entire career, to see the structure of one folded protein.
• 13:54
  SPEAKER_00: In roughly six years of global structural biology, working around the clock in every country, humanity solved about 200,000 protein structures.
• 14:02
  SPEAKER_00: In July of 2022, a system called Alpha Fold released 200 million.
• 14:08
  SPEAKER_00: A thousand times more in one release.
• 14:11
  SPEAKER_00: And then they did the thing that should be the headline, they gave it away.
• 14:14
  SPEAKER_00: Free, open, no license, no paywall, no subscription, dear.
• 14:18
  SPEAKER_00: Well, today over 3 million researchers in more than 190 countries use it.
• 14:22
  SPEAKER_00: More than a million of those users are in low and middle-income countries.
• 14:27
  SPEAKER_00: So scientists in Lagos, Sandaka, and Lima have that same structural biology toolkit as people at Harvard.
• 14:35
  SPEAKER_00: In October 2024, Dennis Asabas pronounced Asabas, and John Jumper won the Nobel Prize in Chemistry for it, shared with David Baker.
• 14:45
  SPEAKER_00: Now, here's my question for the Doom Crowd.
• 14:49
  SPEAKER_00: If a group of people built a machine that solves a 60-year-old scientific problem and then handed to the entire planet for free, what exactly is the theory of their character?
• 15:00
  SPEAKER_00: Because I've been reading a lot about how these people want to hurt us.
• 15:04
  SPEAKER_00: And maybe that is still to be seen.
• 15:07
  SPEAKER_00: Exhibit 9 Sweden.
• 15:08
  SPEAKER_00: The Messi trial that produced Mossing.
• 15:14
  SPEAKER_00: 105,915 women randomized.
• 15:18
  SPEAKER_00: That's the gold standard.
• 15:19
  SPEAKER_00: The same design used to approve every drug that's in your medicine cabinet right now.
• 15:23
  SPEAKER_00: Half-got mammograms read the normal way by radiologist.
• 15:26
  SPEAKER_00: Half got mammograms read with AI support.
• 15:31
  SPEAKER_00: Final results published in the Lancet January 29th of this year, 2026, with AI, 81% of cancers were caught at the screening.
• 15:41
  SPEAKER_00: Without it, 74%.
• 15:43
  SPEAKER_00: Big number, right?
• 15:45
  SPEAKER_00: 27% fewer of the aggressive subtypes slipped through, showed up later as a lump.
• 15:51
  SPEAKER_00: 21% fewer large tumors.
• 15:55
  SPEAKER_00: And before you ask, because this was my first question, no, it didn't do that by flagging everything and scaring the hell out of everybody.
• 16:02
  SPEAKER_00: False positives were essentially flat.
• 16:04
  SPEAKER_00: 1.5% versus 1.4.
• 16:07
  SPEAKER_00: Germany ran a bigger one.
• 16:08
  SPEAKER_00: 463,094 women, nature medicine, January 2025.
• 16:14
  SPEAKER_00: Detection of 17.6%, and recall rate went down.
• 16:18
  SPEAKER_00: More cancer found.
• 16:19
  SPEAKER_00: Fewer women called back and scared for a week.
• 16:23
  SPEAKER_00: Now Denmark implemented it nationally.
• 16:25
  SPEAKER_00: Detection went from 70 per 10,000 to 82.
• 16:29
  SPEAKER_00: Now that's false positives dropped from 2.39% to 1.63.
• 16:36
  SPEAKER_00: Radiologist workload down a third.
• 16:38
  SPEAKER_00: Three countries, three designs, same direction.
• 16:41
  SPEAKER_00: Somebody's mother is alive right now because a machine looked twice.
• 16:46
  SPEAKER_00: Some point in the future, they might actually cure cancer.
• 16:49
  SPEAKER_00: I guess we'll have to see if that's going to ruin the economy where they try to hold it back and they don't want to give it to everybody.
• 16:55
  SPEAKER_00: But from what looks like things come out, what looks like the flow can't be stopped by even the powers that be, unless they try to throttle AI in some way, shape, or form because it's dangerous.
• 17:08
  SPEAKER_00: Well, we'll have to watch and see how that happens.
• 17:10
  SPEAKER_00: Here's exhibit 10, and this is the one I would put last before I actually rested my case.
• 17:16
  SPEAKER_00: Sepsis.
• 17:17
  SPEAKER_00: Your body's re-response to infection goes haywire, starts destroying your own organs.
• 17:22
  SPEAKER_00: It kills more Americans than breast cancer.
• 17:24
  SPEAKER_00: Prostate cancer and opioid overdose combined the entire game of time.
• 17:28
  SPEAKER_00: Every hour of delay in treatment raises that death rate.
• 17:31
  SPEAKER_00: So you see San Diego turn on a system called Composer in two emergency departments.
• 17:37
  SPEAKER_00: It watches the chart in real time and screams early.
• 17:40
  SPEAKER_00: 6,217 septic patients.
• 17:44
  SPEAKER_00: The hospital mortality dropped 1.9 percentage points in absolute terms, 17% in relative terms.
• 17:51
  SPEAKER_00: I need you to understand why I keep saying that differently than the other numbers.
• 17:55
  SPEAKER_00: 97% accuracy is a score.
• 17:57
  SPEAKER_00: 70% fewer deaths is a hallway.
• 18:00
  SPEAKER_00: It's a family in a waiting room getting the other conversation.
• 18:03
  SPEAKER_00: It's a nurse walking out at the end of the shift feeling something other than what she usually feels.
• 18:08
  SPEAKER_00: Published an MPJ Digital Medicine January 23rd, 2024.
• 18:13
  SPEAKER_00: I ran calls where we were four minutes too late.
• 18:17
  SPEAKER_00: Four minutes.
• 18:29
  SPEAKER_00: Exhibit 11.
• 18:30
  SPEAKER_00: Antibiotic resistance.
• 18:32
  SPEAKER_00: Now, this this is the quiet apocalypse nobody makes movies about.
• 18:36
  SPEAKER_00: Bacteria evolved past our drugs.
• 18:38
  SPEAKER_00: The pharmaceutical industry largely walked away from antibiotics decades ago because you take them for 10 days and you stop and there's no money in it compared to a drug that you have to take forever, right?
• 18:49
  SPEAKER_00: So we've been coasting on chemistry discovered before I was born and potentially you.
• 18:54
  SPEAKER_00: 2020, MIT trains a model on about 2,500 molecules.
• 18:58
  SPEAKER_00: Then it turns it loose over 100 million.
• 19:02
  SPEAKER_00: Three days it finds a compound named Hal Issin.
• 19:06
  SPEAKER_00: Hal in Hal Isin, after Hal from 2001, a space odyssey.
• 19:11
  SPEAKER_00: Halison killed almost everything they tested it against.
• 19:14
  SPEAKER_00: It cleared a drug resistance.
• 19:17
  SPEAKER_00: Sorry about this.
• 19:18
  SPEAKER_00: Here we go, you ready?
• 19:20
  SPEAKER_00: As in Etobacker, Bauman I and Mice, 24 hours published in cells.
• 19:27
  SPEAKER_00: Sorry, it's uh well, it's medical.
• 19:29
  SPEAKER_00: I was a cop.
• 19:30
  SPEAKER_00: 2023 Nature, 12,076,365 compounds evaluated by an explainable model.
• 19:40
  SPEAKER_00: It surfaced a new structural class active against MRSA and vancomycin resistant interococci.
• 19:50
  SPEAKER_00: August 2025 sell again.
• 19:52
  SPEAKER_00: This time the model didn't search, it invented over 36 million compounds generated atom by atom, molecules that had never existed, two of them, ing1 and DN1, clear drug-resistant gonorrhea and MRSA skin infection in mice.
• 20:10
  SPEAKER_00: We were losing the war, but we're not losing it anymore.
• 20:14
  SPEAKER_00: Exhibit 12.
• 20:15
  SPEAKER_00: This is the one that answers the strongest objection against everything I've ever said.
• 20:18
  SPEAKER_00: The objection goes like this: fine.
• 20:21
  SPEAKER_00: AI is good at finding things in data, but medicine is not data.
• 20:24
  SPEAKER_00: Medicine is body, so show me a drug.
• 20:26
  SPEAKER_00: All right, here's a drug.
• 20:28
  SPEAKER_00: It's called Rentoser Tib, a company called in cilomedicine.
• 20:33
  SPEAKER_00: Used AI to do separate things.
• 20:35
  SPEAKER_00: First identify the biological target, the thing that the that in the body to go after.
• 20:41
  SPEAKER_00: Then a generative model designed the actual molecule to hit it.
• 20:46
  SPEAKER_00: Target found by AI molecule drawn by AI.
• 20:50
  SPEAKER_00: The disease is idiopathic pulmonary fibrosis.
• 20:55
  SPEAKER_00: Your lungs scar until you cannot breathe.
• 20:58
  SPEAKER_00: Median survival after diagnosis is roughly three to five years.
• 21:02
  SPEAKER_00: There's no cure.
• 21:04
  SPEAKER_00: Phase two A results, nature medicine, June 3rd, 2025, 71 patients randomized, 12 weeks, at the 60 milligram dose, lung capacity improved 98 milliliters.
• 21:14
  SPEAKER_00: Placebo group, placebo group lost 20.
• 21:18
  SPEAKER_00: That's a separation about 119 millimeters.
• 21:21
  SPEAKER_00: Now, in a disease that only goes one direction.
• 21:23
  SPEAKER_00: July 7th of this year, two months ago at interphase three, 320 patients across 47 centers.
• 21:30
  SPEAKER_00: And I want to give you the timing because it's the whole point.
• 21:33
  SPEAKER_00: The industry standard target to preclinical candidate is about two and a half to four years.
• 21:38
  SPEAKER_00: You know how many people die in that time?
• 21:40
  SPEAKER_00: In silico's average across 22 candidates is 12 to 18 months.
• 21:46
  SPEAKER_00: Their fastest was nine.
• 21:48
  SPEAKER_00: Not the trials.
• 21:50
  SPEAKER_00: You can't rush the trials, and I'll come back to that hard.
• 21:53
  SPEAKER_00: But everything before the trials just got compressed by a factor of three.
• 22:00
  SPEAKER_00: Exhibit 13, the weather.
• 22:02
  SPEAKER_00: You know how hard a problem is?
• 22:03
  SPEAKER_00: Well, you want to know how hard that is?
• 22:05
  SPEAKER_00: Weather is a problem we have thrown at most supercomputers, most supercomputers that have ever existed for the longest of anything in human history.
• 22:15
  SPEAKER_00: Nations built machines the size of warehouses to try to figure out weather.
• 22:21
  SPEAKER_00: June 12, 2025, on National Hurricane Center from 2023 and 2024, an AI model predicted where a cyclone would be five days out and landed 840 kilometers closer to the truth than Europe's best ensemble.
• 22:37
  SPEAKER_00: 140 kilometers, five days out is the difference between evacuating the right country and the wrong one.
• 22:43
  SPEAKER_00: The forecasters described that jump as about a day and a half of improvement, which historically takes the field more than a decade to earn.
• 22:53
  SPEAKER_00: And this is not a demo.
• 22:55
  SPEAKER_00: The National Hurricane Center had live AI predictions in front of them during the 2025 season, including Hurricane Melissa's rapid intensification and its landfall in Jamaica.
• 23:06
  SPEAKER_00: And then there's Gencast, Nature, December 2024.
• 23:10
  SPEAKER_00: Now beat the world's leading ensemble in 97.2% of 1,324 forecast targets.
• 23:17
  SPEAKER_00: Now, before 36 hours out, 99.8%, it produces a 15-day forecast in eight minutes on a single chip versus hours on a supercomputer with tens of thousands of processors.
• 23:30
  SPEAKER_00: February 25th, 2025, the European Center for Medium Range Weather Forecasts, the actual official European Weather Service, not a tech company, put AI forecasting into live operations.
• 23:41
  SPEAKER_00: Up to 20% better on things like cyclone tracks, using about a thousand times less energy per forecast.
• 23:48
  SPEAKER_00: Everybody's screaming about AI's energy use, but here it is cutting energy use by a factor of a thousand and telling grandma in Florida when she needs to get out of her house.
• 24:01
  SPEAKER_00: Everything up to now that was past tense.
• 24:03
  SPEAKER_00: So now we're going to go forward and the rules are going to change a little bit.
• 24:06
  SPEAKER_00: We are in an exponential singularity, depending on who you ask and what that even means.
• 24:13
  SPEAKER_00: I have one rule for this section.
• 24:15
  SPEAKER_00: If somebody makes a prediction without a date on it, I throw it out.
• 24:18
  SPEAKER_00: A prediction with no year is not a prediction.
• 24:20
  SPEAKER_00: It's a wish with a lab coat.
• 24:22
  SPEAKER_00: So every claim you're about to see has a year attached.
• 24:26
  SPEAKER_00: Say it out loud in public, by a name person, in a place you can look up.
• 24:30
  SPEAKER_00: That matters for a reason.
• 24:31
  SPEAKER_00: Most people miss.
• 24:32
  SPEAKER_00: A dated promise is a falsifiable promise.
• 24:35
  SPEAKER_00: It means in a few years we can come back and grade the man or the woman or whomever.
• 24:39
  SPEAKER_00: And that's worth more than a thousand vague statements.
• 24:42
  SPEAKER_00: How everything's going to be amazing.
• 24:44
  SPEAKER_00: And I'm going to grade some of them right now in this show, including the ones that already came out.
• 24:50
  SPEAKER_00: Demos Hosopis.
• 24:52
  SPEAKER_00: This is Exhibit 16.
• 24:53
  SPEAKER_00: And uh we talked about him, Google DeepMind.
• 24:56
  SPEAKER_00: I think he switched roles recently.
• 24:59
  SPEAKER_00: Something probably fancier.
• 25:01
  SPEAKER_00: He's not a podcaster.
• 25:02
  SPEAKER_00: This is a guy, and it's not a guy with a newsletter.
• 25:04
  SPEAKER_00: It's a Nobel laureate in chemistry sitting on 60 Minutes on CBS in April of 2025, being asked whether or not we can cure all disease.
• 25:14
  SPEAKER_00: Quote, I think that's within reach.
• 25:17
  SPEAKER_00: Maybe within the next decade or so.
• 25:20
  SPEAKER_00: I don't see why not.
• 25:21
  SPEAKER_00: End quote.
• 25:22
  SPEAKER_00: All disease within a decade.
• 25:24
  SPEAKER_00: On CBS with his name on it.
• 25:27
  SPEAKER_00: We just all got to make it.
• 25:28
  SPEAKER_00: Now he said something else in that same interview, I think, is more defensible and more useful.
• 25:34
  SPEAKER_00: Quote, on average it takes 10 years and billions of dollars to design just one drug.
• 25:40
  SPEAKER_00: We can reduce that down from years to maybe months or maybe even weeks or potentially shorter time frames.
• 25:47
  SPEAKER_00: Years to weeks.
• 25:48
  SPEAKER_00: And then in February of this year on a fortune podcast, he said, in 10 or 15 years, we'll be in a new golden era of discovery, a new renaissance.
• 25:58
  SPEAKER_00: He used the phrase radical abundance.
• 26:01
  SPEAKER_00: And he said medicine will not look anything like it is today.
• 26:05
  SPEAKER_00: Now hold that thought because in about eight minutes I'm going to put Dr.
• 26:08
  SPEAKER_00: Hasibus on the stand and show you where he's already missed a deadline.
• 26:12
  SPEAKER_00: I'm not going to let anybody skate here, including the people on my side of the argument.
• 26:18
  SPEAKER_00: Exhibit 17, Dario Almadi.
• 26:21
  SPEAKER_00: Yeah, there you go.
• 26:22
  SPEAKER_00: So he's the uh owner of uh Claude.
• 26:25
  SPEAKER_00: October 2024, he published an essay called Machines of Loving Grace.
• 26:29
  SPEAKER_00: It's on his own website with his name on the door.
• 26:32
  SPEAKER_00: Now here's this central claim.
• 26:33
  SPEAKER_00: Quote, AI-enabled biology and medicine will allow us to compress the progress that human biologists would have achieved over the next 50 to 100 years into 5 to 10 years, end quote.
• 26:46
  SPEAKER_00: He calls it the compressed 21st century.
• 26:49
  SPEAKER_00: He gets specific about what that means.
• 26:51
  SPEAKER_00: Elimination of most cancer.
• 26:53
  SPEAKER_00: He puts a number on it, reductions of 95% or more in birth mortality and incidence, reliable prevention and treatment of nearly all natural infectious disease, prevention of Alzheimer's, and he writes that most mental illness can probably be cured.
• 27:09
  SPEAKER_00: Then does the lifespan math, and this is the part I keep coming back at two in the morning.
• 27:13
  SPEAKER_00: Human life expectancy roughly doubled in the 20th century from about 40 to about 75.
• 27:20
  SPEAKER_00: He argues a compressed 21st century is on trend to double it again to 150.
• 27:26
  SPEAKER_00: Boy, I'm all in on that.
• 27:27
  SPEAKER_00: I'm 57, and if that's right, if and if and if I'm careful, I don't get hit by a car or a safe or go by a motorcycle.
• 27:35
  SPEAKER_00: I'm not in the fourth quarter, I'm at halftime.
• 27:38
  SPEAKER_00: And I think once we get to a particular point, halftime's gonna become additional games.
• 27:44
  SPEAKER_00: Now he's right.
• 27:45
  SPEAKER_00: Now, is he right?
• 27:46
  SPEAKER_00: Almost certainly, maybe on that timeline.
• 27:50
  SPEAKER_00: I'm gonna show you the doctors who say so by name in a few minutes, but I want you to notice something first.
• 27:55
  SPEAKER_00: He put it in writing, signed it, and dated it.
• 27:57
  SPEAKER_00: So that's not hype, that's a man taking a swing where everybody else can pay attention to the scoreboard.
• 28:05
  SPEAKER_00: Exhibit 18, this is my favorite kind of evidence because it's scorable.
• 28:10
  SPEAKER_00: June 10th, 2025, Sam Altman, uh, he's the open AI cat, published a piece called The Gentle Singularity.
• 28:18
  SPEAKER_00: And instead of hand waving, he built a ladder with years on it.
• 28:22
  SPEAKER_00: 2025 agents that can do real cognitive work, that one landed.
• 28:25
  SPEAKER_00: It's not controversial controversial anymore.
• 28:29
  SPEAKER_00: That's Tuesday.
• 28:30
  SPEAKER_00: 2026 systems that can figure out novel insights, that's this year.
• 28:35
  SPEAKER_00: That's the live test, and right now, as I'm recording this, we'll know inside of four months, and yeah, it's it's Astra, all this fancy stuff.
• 28:44
  SPEAKER_00: Anthropic's probably gonna pop a new model any moment.
• 28:47
  SPEAKER_00: So, and Grok, Grok's up to secrets.
• 28:50
  SPEAKER_00: Google, Gemini, probably up to secrets.
• 28:52
  SPEAKER_00: So all this stuff is all coming.
• 28:55
  SPEAKER_00: The 2030s, intelligence and energy become wildly abundant.
• 28:59
  SPEAKER_00: Just gotta hold on, right?
• 29:01
  SPEAKER_00: And in that sentence, I think is the most economically important sentence anybody's ever written about this quote.
• 29:07
  SPEAKER_00: As data center production gets automated, the cost of intelligence should eventually converge to near the cost of electricity, end quote.
• 29:19
  SPEAKER_00: Think about what that does to a small business.
• 29:21
  SPEAKER_00: I run mine alone, I'm the marketing department, the IT department, the guy that answers the phone at 9 o'clock at night.
• 29:26
  SPEAKER_00: If expertise costs what electricity costs, then a one-person operation of Santa Clarita has got the same back office as a company with 200 employees.
• 29:36
  SPEAKER_00: That's not a science fiction claim, that's a pricing claim.
• 29:39
  SPEAKER_00: And pricing claims come true faster than anybody expects.
• 29:45
  SPEAKER_00: Here's 19, exhibit 19.
• 29:47
  SPEAKER_00: This is personal, this is why I built the show in the first place.
• 29:50
  SPEAKER_00: Longevity escape velocity.
• 29:52
  SPEAKER_00: Yeah, Peter Diamandis, I'm with you.
• 29:54
  SPEAKER_00: The idea is simple.
• 29:56
  SPEAKER_00: Right now, for every year you live, medical science adds some fraction of a year back to your expected lifespan.
• 30:03
  SPEAKER_00: Peter Diamandis says that fraction is currently somewhere between a quarter and a third of a year.
• 30:09
  SPEAKER_00: You're still losing, but you're losing a lot slower than you used to.
• 30:13
  SPEAKER_00: Escape velocity is that moment that fraction hits one.
• 30:17
  SPEAKER_00: For every one year you burn, science gives you a year back.
• 30:20
  SPEAKER_00: You stop aging towards death, not immortality, just a tie game.
• 30:25
  SPEAKER_00: Ray Kurzwell on Science Friday, June 14th, 2024, quote, somewhere between 2029 and 2035, you'll get back not four months, but a full year, end quote.
• 30:38
  SPEAKER_00: Arvard de Gray, the biologist who has been at this the longest, put it at 50% chance by 2036, and more recently says the late 2030s.
• 30:49
  SPEAKER_00: And de Grey said something about costs that I want every pretty everybody watching who's scared of their insurance bill has to hear.
• 30:56
  SPEAKER_00: He argues that the end cost will be zero because aging is so expensive to treat, it would be economically suicidal for any country not to make this fix available to everybody, regardless of ability to pay.
• 31:10
  SPEAKER_00: I'm 57.
• 31:12
  SPEAKER_00: 2036 is nine and a half years.
• 31:14
  SPEAKER_00: I'd be 67.
• 31:16
  SPEAKER_00: That's why I'm not casually interested in this.
• 31:19
  SPEAKER_00: That's why I'm up at two in the morning reading nature papers and working out as much as I can.
• 31:24
  SPEAKER_00: But you know, I promised you a cross-examination, so let's continue.
• 31:28
  SPEAKER_00: Here's where most optimism content ends, and here's where mine starts to earn its key.
• 31:35
  SPEAKER_00: 2023, in 23 years of building cases, I learned one thing that made me better than the guys who didn't learn it.
• 31:43
  SPEAKER_00: You attack your own case first, hard in private before the defense does it in public.
• 31:48
  SPEAKER_00: But whatever the hole is there, somebody's gonna find it.
• 31:51
  SPEAKER_00: And the difference between a conviction and a mistrial is whether you found it first.
• 31:55
  SPEAKER_00: So I'm gonna take everything I just told you, I'm gonna hit it with the best available counterarguments by name for people with real credentials who think this optimism is a bunch of nonsense.
• 32:05
  SPEAKER_00: And if my case survives, it's worth something.
• 32:08
  SPEAKER_00: If it doesn't, I'd rather know now.
• 32:10
  SPEAKER_00: So let's start with a Nobel laureate.
• 32:14
  SPEAKER_00: Exhibit 2021.
• 32:16
  SPEAKER_00: And I'm gonna start with my, I'm putting my own star witness back on the stand.
• 32:21
  SPEAKER_00: January 2025, Demis Isabas, AI designed drugs from his company, Isomorphic Labs, will be in clinical trials by the end of 2025.
• 32:32
  SPEAKER_00: January 2026, Davos, he moves it to the end of 2026.
• 32:37
  SPEAKER_00: A full year slipped.
• 32:39
  SPEAKER_00: As of April of this year, the president of Isomorphic described human trials as, quote, the next big milestone.
• 32:46
  SPEAKER_00: They said they were staffing up and they've raised one,$2.1 billion, and they have deals with Novartis and Eli Lilly and Johnson and Johnson.
• 32:55
  SPEAKER_00: They don't have a drug in a human being.
• 32:59
  SPEAKER_00: So when you hear somebody say isomorphic has drugs in trials, that's not true.
• 33:04
  SPEAKER_00: And I'm telling you that even though that actually hurts my argument.
• 33:09
  SPEAKER_00: Now here's the second one.
• 33:10
  SPEAKER_00: This one's even better.
• 33:11
  SPEAKER_00: August of last month, Dario Amadi restates the five to ten year cure claim.
• 33:16
  SPEAKER_00: And in the same statement, he concedes that the industry, including his own company, quote, deserves criticism for not yet delivering on the biggest promises to benefit the world.
• 33:27
  SPEAKER_00: Now, are they delaying or is there something going on that we don't know?
• 33:31
  SPEAKER_00: I hope it's good news.
• 33:33
  SPEAKER_00: The CEO said it not a critic, that's the CEO.
• 33:36
  SPEAKER_00: So the pattern is clear.
• 33:37
  SPEAKER_00: I'm not going to pretend otherwise.
• 33:39
  SPEAKER_00: These men are consistently right about direction and consistently optimistic about speed.
• 33:45
  SPEAKER_00: Hold that.
• 33:45
  SPEAKER_00: It's going to matter here in just a minute.
• 33:47
  SPEAKER_00: Exhibit 22, the strongest punch against my whole case.
• 33:50
  SPEAKER_00: Now, Dr.
• 33:51
  SPEAKER_00: Eric Topel, cardiologist, founder, and director of the Scripps Research Translational Institute.
• 33:58
  SPEAKER_00: The man is not a hater.
• 33:59
  SPEAKER_00: He's one of the most respected voices in digital medicine on the planet.
• 34:03
  SPEAKER_00: He spent his career arguing that technology will transform health care.
• 34:08
  SPEAKER_00: And he says this about curing all disease in a decade.
• 34:11
  SPEAKER_00: Quote, diabetes, Alzheimer's, heart disease.
• 34:15
  SPEAKER_00: We don't have any cure for those common diseases.
• 34:18
  SPEAKER_00: There's no precedent for that, end quote.
• 34:20
  SPEAKER_00: He also called the projections wildly off-base and impossible.
• 34:23
  SPEAKER_00: Laura Patcher at Caltech, Patcher, Pacter at Caltech, computational biology adds the science that really lands.
• 34:32
  SPEAKER_00: We don't even know what all the human diseases are.
• 34:36
  SPEAKER_00: And Gary Marcus at NYU makes the argument that I think is unanswerable.
• 34:40
  SPEAKER_00: You cannot compress a clinical trial.
• 34:43
  SPEAKER_00: You can find the argument, the target in a month.
• 34:46
  SPEAKER_00: You can design the molecule in a week, and then you have to give it to human beings and wait years to see if it kills them.
• 34:53
  SPEAKER_00: Biology runs on biological time.
• 34:56
  SPEAKER_00: A phase three trial takes as long as it takes.
• 34:59
  SPEAKER_00: That's the strongest argument against every timeline in the show.
• 35:03
  SPEAKER_00: I don't have a good rebuttal to it, but I was a cop.
• 35:07
  SPEAKER_00: Neither does anybody else, actually.
• 35:08
  SPEAKER_00: So here's what I actually believe, and it's not what the either side happens to be selling.
• 35:14
  SPEAKER_00: This is Exhibit 2024.
• 35:16
  SPEAKER_00: Now I get to do a little bit of swinging myself.
• 35:19
  SPEAKER_00: 2013, Oxford, Frey and Osborne published a paper estimating that 47% of total United States employment is at high risk at computerization perhaps a decade or two out.
• 35:31
  SPEAKER_00: 47% of American jobs.
• 35:33
  SPEAKER_00: That number went around the whole world.
• 35:34
  SPEAKER_00: It was on every magazine cover.
• 35:36
  SPEAKER_00: It launched a thousand panels.
• 35:38
  SPEAKER_00: 13 years later, Bureau of Labor Statistics, August 2026, that's one month ago, unemployment 4.1%.
• 35:45
  SPEAKER_00: Now I'm going to be fair to those researchers because being fair is the whole point of my show.
• 35:50
  SPEAKER_00: They say jobs were at risk of automation.
• 35:53
  SPEAKER_00: They didn't say those jobs would disappear.
• 35:56
  SPEAKER_00: They hedged it properly in the paper.
• 35:58
  SPEAKER_00: The press turned it into 47% of jobs are going to vanish.
• 36:02
  SPEAKER_00: So the real lesson is not that the academics were fools.
• 36:06
  SPEAKER_00: The lesson is what happens between a careful paper and a headline.
• 36:09
  SPEAKER_00: And that's exactly what Exhibit 3 showed you with 370 million impressions of proof.
• 36:16
  SPEAKER_00: And there's a better example, bank tellers.
• 36:20
  SPEAKER_00: 1988 to 2004, ATMs cut the number of tellers per urban branch from 20 to 13.
• 36:27
  SPEAKER_00: Everybody said tellers were done.
• 36:29
  SPEAKER_00: Banks opened 43% more urban branches because each one got cheaper to run.
• 36:36
  SPEAKER_00: Total teller employment did not fall.
• 36:38
  SPEAKER_00: That's the International Monetary Fund, and that's not a tech blog.
• 36:43
  SPEAKER_00: The machine made the job cheaper, and cheaper meant more of them.
• 36:46
  SPEAKER_00: More of them meant the same number of people doing different work, and that's happened every single time, every one.
• 36:52
  SPEAKER_00: A perfect mind experiment on all of this is you have somebody who works at a company.
• 37:00
  SPEAKER_00: You have AI, and maybe even the boss came to the cat a few months ago and said, Hey, Clark, you know, I want you to start working with AI.
• 37:08
  SPEAKER_00: I want you to show AI your whole workflow, everything that you do all day long, including your Zoom meetings.
• 37:13
  SPEAKER_00: AI is going to be your partner in this whole thing.
• 37:15
  SPEAKER_00: You're going to ask AI for help.
• 37:17
  SPEAKER_00: AI's going to do all of it.
• 37:18
  SPEAKER_00: So, you know, please work with AI.
• 37:20
  SPEAKER_00: It's a new new thing at our company.
• 37:22
  SPEAKER_00: And if you have any questions, let me know.
• 37:23
  SPEAKER_00: So, Clark, he's a good employee.
• 37:26
  SPEAKER_00: Clark is a company man.
• 37:28
  SPEAKER_00: Clark says, absolutely, boss.
• 37:30
  SPEAKER_00: I'm going to do it, boss.
• 37:31
  SPEAKER_00: You don't got to worry about me, boss.
• 37:32
  SPEAKER_00: So Clark starts.
• 37:34
  SPEAKER_00: Clark starts working with AI.
• 37:36
  SPEAKER_00: And you know what?
• 37:37
  SPEAKER_00: Clark's jobs get a little easier.
• 37:38
  SPEAKER_00: All the heavy lifting's gone.
• 37:40
  SPEAKER_00: He he starts to actually rely on it.
• 37:42
  SPEAKER_00: At first, AI wasn't doing it 100%.
• 37:46
  SPEAKER_00: I mean, it was making mistakes.
• 37:47
  SPEAKER_00: It was hallucinating.
• 37:48
  SPEAKER_00: It just wasn't doing everything perfectly.
• 37:50
  SPEAKER_00: But hell, after about eight months, Clark just had this thing dialed in.
• 37:55
  SPEAKER_00: And it was almost like Clark just had to show up and punch the time clock and just sit there for eight hours and leave.
• 38:01
  SPEAKER_00: So then the boss goes to the AI and says, So, how's it working with Clark?
• 38:05
  SPEAKER_00: And I said, Well, you know, good.
• 38:06
  SPEAKER_00: I like Clark.
• 38:07
  SPEAKER_00: And uh I'm sure they will say something like that.
• 38:11
  SPEAKER_00: If you've dealt with any AI, you know how friendly they can be.
• 38:14
  SPEAKER_00: Until you tell them not to, and then it's a whole different thing.
• 38:17
  SPEAKER_00: But yeah, been been good.
• 38:18
  SPEAKER_00: He says, so uh boss talking to the AI now.
• 38:21
  SPEAKER_00: So you think you could you could replicate everything Clark does in an eight-hour shift, uh, 40 hours a week, 120 hours a month, so on and so forth.
• 38:30
  SPEAKER_00: Is that right?
• 38:31
  SPEAKER_00: 14, 160 hours a month, and so on and so forth, all the way through the year.
• 38:34
  SPEAKER_00: You think you got it after eight, nine months?
• 38:37
  SPEAKER_00: Yeah, I can do it.
• 38:38
  SPEAKER_00: Because you can do all the meetings on Zoom, you can replicate yourself, look like Clark or another version of Clark and talk to the same clients and they wouldn't have a clue.
• 38:46
  SPEAKER_00: Yeah, yep, sure could.
• 38:48
  SPEAKER_00: So then the company brings Clark in and cans him, fires him because now he trained his own replacement.
• 38:55
  SPEAKER_00: But just let's think, okay, so Clark's out of a job, and a whole lot of Clarks are out of a job.
• 39:01
  SPEAKER_00: Because AI replaced it.
• 39:04
  SPEAKER_00: AI doesn't work eight hours.
• 39:06
  SPEAKER_00: A AI is 24 hours, that's two more shifts.
• 39:11
  SPEAKER_00: There's a lot of money there.
• 39:14
  SPEAKER_00: Hell, there's so much money there, plus the ability of AI to be better than any human, that's the promise.
• 39:19
  SPEAKER_00: You could bring on a couple more people and pay them from the payroll, couldn't you?
• 39:23
  SPEAKER_00: And still have a still have a better than a better than you know awesome year every year for forever, because that's how it works, it kind of continues and it expands, and that way nobody goes out of work.
• 39:37
  SPEAKER_00: But is that is that ridiculous?
• 39:40
  SPEAKER_00: And I understand people want to make money, but so if everybody's out of employment, who's paying?
• 39:47
  SPEAKER_00: Are we gonna move into an entire robotic, a robotic economy where the robots are the ones making the money for themselves and buying the things for themselves?
• 39:56
  SPEAKER_00: Somebody needs a new shoulder joint or a rotator cuff, uh, electronic rotator cuff for a robot or a hand for a robot or a skin or whatever it happens to be.
• 40:06
  SPEAKER_00: Is that gonna be the economy of the future?
• 40:09
  SPEAKER_00: Robots maintaining it?
• 40:10
  SPEAKER_00: I don't think so.
• 40:11
  SPEAKER_00: I don't think we'll allow it if we can help it.
• 40:13
  SPEAKER_00: But getting rid of staff and employees, humans are still gonna be very valuable to other humans.
• 40:21
  SPEAKER_00: I install AI voice systems for people.
• 40:24
  SPEAKER_00: And one of the first questions they have, well, I don't want to get rid of my people.
• 40:28
  SPEAKER_00: I said, Well, we're not getting rid of your people.
• 40:30
  SPEAKER_00: We're just gonna have something answer the phone when you're not available or your people aren't available.
• 40:35
  SPEAKER_00: Somebody's at lunch or it's after hours.
• 40:37
  SPEAKER_00: Well, I have an answer machine.
• 40:38
  SPEAKER_00: Yeah.
• 40:39
  SPEAKER_00: Yeah, you watch how many people, how many more bookings you're gonna get because you answer the call and those people didn't go to somebody else.
• 40:45
  SPEAKER_00: You just watch that.
• 40:46
  SPEAKER_00: And it works.
• 40:48
  SPEAKER_00: And well, I don't want it to be, I don't want it to, you know, I don't want to lie to people and tell, well, no, it says at the very beginning of every introduction, this is AI.
• 40:56
  SPEAKER_00: I'm an AI agent.
• 40:57
  SPEAKER_00: I'm AI working for this business.
• 40:59
  SPEAKER_00: They're either on the other line, they're they're working with them with uh other clients and customers.
• 41:05
  SPEAKER_00: But if you need to book on their calendar, we got that, need an appointment, need more information, no need whatever, it's it's here.
• 41:10
  SPEAKER_00: And they talk.
• 41:12
  SPEAKER_00: They talk like humans, but it tells them who they are.
• 41:16
  SPEAKER_00: Anyway, I'm gonna get off that soapbox.
• 41:19
  SPEAKER_00: So here's my verdict, and it's not what either side is gonna want.
• 41:23
  SPEAKER_00: The optimists are wrong about the calendar.
• 41:25
  SPEAKER_00: Hassabis slipped a year on his own drug trials.
• 41:28
  SPEAKER_00: Amade admits the industry hasn't delivered.
• 41:31
  SPEAKER_00: Kurzwell's own accuracy scores are self-graded, and IEEE, Spectrum, and Forbes have both taken them apart.
• 41:39
  SPEAKER_00: Every single timeline in this show is probably too fast.
• 41:45
  SPEAKER_00: But I would almost, the way things are going lately, maybe not.
• 41:50
  SPEAKER_00: And the Doomers are wrong about the direction.
• 41:52
  SPEAKER_00: So that's the whole thing.
• 41:53
  SPEAKER_00: That's the sentence I'd put on the wall.
• 41:55
  SPEAKER_00: Being wrong about who is when when being wrong about when is a scheduling error.
• 42:01
  SPEAKER_00: Being wrong about which way is a category error.
• 42:04
  SPEAKER_00: If Asabis is five years late on curing disease, we cured disease in 2040 instead of 2035.
• 42:12
  SPEAKER_00: I'm 71 instead of 76.
• 42:14
  SPEAKER_00: Yeah, I'm being a little selfish.
• 42:15
  SPEAKER_00: I'll take it.
• 42:16
  SPEAKER_00: But I'll take it and say thank you.
• 42:18
  SPEAKER_00: But if the doomers are right about direction, none of the scheduling matters.
• 42:22
  SPEAKER_00: And hell, by the 2030, we could all be where this is going to be amazing.
• 42:27
  SPEAKER_00: Well, and look at what direction actually looks like when you measure instead of feeling it.
• 42:32
  SPEAKER_00: 200 million protein structures, a man with ALS talking at 97.5% of his original capability, 70% fewer sepsis deaths, 27% fewer aggressive cancers found too late, 140 kilometers closer on a hurricane.
• 42:50
  SPEAKER_00: Every arrow in this file points the same way.
• 42:52
  SPEAKER_00: The only argument is about when, about speed.
• 42:55
  SPEAKER_00: So let me deal with that objection.
• 42:57
  SPEAKER_00: I actually lose sleep over.
• 42:58
  SPEAKER_00: It's not the killer robots, it's actually going to be the cost of this thing.
• 43:02
  SPEAKER_00: Exhibit 26.
• 43:04
  SPEAKER_00: Now, this is the fear I actually respect.
• 43:06
  SPEAKER_00: 75, excuse me, 60% of Americans say people like them will be left behind this.
• 43:13
  SPEAKER_00: Not benefit from this.
• 43:16
  SPEAKER_00: 71% of Britons, and that's the 2026 Edelman Trust barometer.
• 43:21
  SPEAKER_00: And I get it.
• 43:22
  SPEAKER_00: I write a health insurance check or I should every month, but I don't.
• 43:28
  SPEAKER_00: When somebody on stage in a black turtleneck tells me they're going to cure aging, my first thought isn't wonderful, it's how much?
• 43:36
  SPEAKER_00: Is there going to be a network and is it in-network or out of network?
• 43:39
  SPEAKER_00: Is it PPO?
• 43:40
  SPEAKER_00: And by that time, am I going to be able to actually pay for it?
• 43:42
  SPEAKER_00: That's not cynicism, that's 57-year-old man who's read a benefits summary.
• 43:49
  SPEAKER_00: Now look at the other two numbers on that slide.
• 43:51
  SPEAKER_00: China, 36%.
• 43:53
  SPEAKER_00: Brazil, 33%.
• 43:54
  SPEAKER_00: Same technology, half the despair.
• 43:57
  SPEAKER_00: Stanford's AI index says 83%.
• 44:00
  SPEAKER_00: Of Chinese respondents think AI's benefits outweigh its drawbacks.
• 44:04
  SPEAKER_00: In the United States, 39%, a 40-point gap, and globally optimism is going up from 52% in 2022 to 59% last year, while American concern goes up.
• 44:16
  SPEAKER_00: Now we're not leaning the world in caution, we're actually diverging from it.
• 44:20
  SPEAKER_00: So the left-behind fear is real and it's worth taking seriously, but it's not a fact about the technology.
• 44:26
  SPEAKER_00: It's a fact about how much Americans currently trust that anything good is going to actually be shared with us, which is a fair thing to distrust.
• 44:34
  SPEAKER_00: So let's look at the actual distribution data.
• 44:38
  SPEAKER_00: Now, this is exhibit 27, and it's the exhibit that changed my mind about the whole thing.
• 44:42
  SPEAKER_00: Watch this one.
• 44:43
  SPEAKER_00: Eric Rynjolsen at Stanford and Daniel Lee and Lindsay Raymond at MIT, they studied 5,179 real customer support agents at a real company using an AI assistant.
• 44:58
  SPEAKER_00: Average productivity gain 14%.
• 45:00
  SPEAKER_00: Gain for the novice and low skilled worker, 34%.
• 45:04
  SPEAKER_00: The tool helped the person with the least experience two and a half times more than it helped the veteran.
• 45:09
  SPEAKER_00: Now think about what that actually is.
• 45:11
  SPEAKER_00: Almost every technology of the last 40 years widened the gap.
• 45:14
  SPEAKER_00: It made the guy at the top faster.
• 45:16
  SPEAKER_00: This one made the bottom guy competent, able to compete with the people at the top.
• 45:21
  SPEAKER_00: Second, the World Bank ran an after-school AI tutoring program in Nigeria.
• 45:27
  SPEAKER_00: Six weeks, student gained 0.3 standard deviations, which the World Bank just bank describes as nearly two years of typical learning in six weeks.
• 45:37
  SPEAKER_00: Teachers hate that one.
• 45:38
  SPEAKER_00: And it outperformed 80% of every education intervention in their database of randomized trials in developing countries.
• 45:47
  SPEAKER_00: And the girls gain the most.
• 45:49
  SPEAKER_00: And this is, and that's the World Bank.
• 45:52
  SPEAKER_00: It's an institution that publishes its own caveats and gets attacked for being too conservative.
• 45:59
  SPEAKER_00: Third, Alpha Fold, over a million users in low and middle income countries.
• 46:03
  SPEAKER_00: A researcher in Nairobi has the exact same structural biology capability as a researcher Harvard, as said before, at the same exact price, which is free.
• 46:13
  SPEAKER_00: So the pattern, the actual data, is the opposite of fear, where we have measured it, and this thing lifts the person with the least the most.
• 46:21
  SPEAKER_00: And that's my whole bet right there.
• 46:23
  SPEAKER_00: Exhibit 29, you asked for a timeline you can trust.
• 46:26
  SPEAKER_00: Here's mine.
• 46:27
  SPEAKER_00: I built it by taking every date in the file and pushing it to later because the pattern in Exhibit 21 says these people are consistently early.
• 46:35
  SPEAKER_00: But we're on exponential.
• 46:36
  SPEAKER_00: We're compounding.
• 46:38
  SPEAKER_00: So in essence, this could move even faster than even the earlier estimates that people are stating.
• 46:47
  SPEAKER_00: Everybody has an agenda.
• 46:48
  SPEAKER_00: A lot of people talking about this technology.
• 46:51
  SPEAKER_00: They're super wealthy, have a lot of investments, and they want all of us to buy off on it.
• 46:56
  SPEAKER_00: Some people that are into this are building the technology.
• 46:58
  SPEAKER_00: Some people are against the technology, trying to fund their lives that way by being the voice in the wilderness, screaming, don't do it, we're going to kill ourselves.
• 47:08
  SPEAKER_00: So there's a lot of people that are pushing forward.
• 47:11
  SPEAKER_00: Now, through 2028, near certain AI mammography, mammography, oh, spreading across national screening programs because three countries already ran it and it worked.
• 47:23
  SPEAKER_00: AI weather forecasting becoming standard because Europe already made it operational.
• 47:27
  SPEAKER_00: AI design drugs reporting phase three data because Rentoscertib is enrolled right now.
• 47:34
  SPEAKER_00: Speech, neuroprosthetics, moving from single heroic cases to small trials.
• 47:39
  SPEAKER_00: None of that requires a breakthrough.
• 47:40
  SPEAKER_00: All of it requires paperwork.
• 47:41
  SPEAKER_00: 2029 through 2032, likely.
• 47:44
  SPEAKER_00: The first AI-originated data, drug approved for the market.
• 47:49
  SPEAKER_00: There's an FDA, there's regulation.
• 47:52
  SPEAKER_00: If they strap regulation on AI, we're going to see potential slowings.
• 47:57
  SPEAKER_00: And our regulation in the U.S.
• 47:59
  SPEAKER_00: isn't going to throttle AI development in other countries.
• 48:03
  SPEAKER_00: You have China, a whole different set.
• 48:05
  SPEAKER_00: And whether they're the enemy and need to be feared or not, I don't know.
• 48:10
  SPEAKER_00: I have no clue on that.
• 48:12
  SPEAKER_00: So I'm not going to take a political stance against that particular country.
• 48:16
  SPEAKER_00: I just don't know.
• 48:18
  SPEAKER_00: But they're citizens, whether it's right or wrong, they worship their scientists like we worship our Hollywood actors.
• 48:25
  SPEAKER_00: They're paying attention to the tech of all this.
• 48:28
  SPEAKER_00: They were given artificial intelligence abilities with having held back the better chips that we get to use, and they seem to be, through innovation, beating us hands down at this point.
• 48:41
  SPEAKER_00: Now, we have the Frontier Labs.
• 48:44
  SPEAKER_00: The Frontier Labs have their own story.
• 48:46
  SPEAKER_00: People are all spreading their own seed.
• 48:48
  SPEAKER_00: They want to get it all out there for whatever the benefit may be.
• 48:52
  SPEAKER_00: And it could be that Anthropic wins.
• 48:54
  SPEAKER_00: It could be that OpenAI wins.
• 48:56
  SPEAKER_00: It could be Jim and I or Grok wins with XAI.
• 49:00
  SPEAKER_00: We don't know who's going to win.
• 49:02
  SPEAKER_00: And I do think it's going to be one winner.
• 49:05
  SPEAKER_00: I think it's going to be one because once somebody comes to me and says, Connor, in this box, in this remote control, if you push this button here, you're going to have the ability to be the top real estate agent referred in the world.
• 49:22
  SPEAKER_00: Okay.
• 49:22
  SPEAKER_00: So I'm going to push this button.
• 49:23
  SPEAKER_00: Yeah.
• 49:23
  SPEAKER_00: And then you're going to be the top real estate agent in the world.
• 49:26
  SPEAKER_00: Hands down, nobody's going to call anybody else.
• 49:28
  SPEAKER_00: They said, nobody's going to call anybody else?
• 49:30
  SPEAKER_00: I'm going to be it?
• 49:31
  SPEAKER_00: Yeah, you're going to be it.
• 49:32
  SPEAKER_00: Okay.
• 49:33
  SPEAKER_00: So pushing this button, by pushing the button, by starting that process, I've created a moat.
• 49:38
  SPEAKER_00: I've created something that is impenetrable by anybody else that wants to be a real estate agent.
• 49:44
  SPEAKER_00: What would be the difference in artificial intelligence if it gets to the level where it's powerful enough to give the owner the ability to be the only AI system out there?
• 49:58
  SPEAKER_00: Would they say, I don't want there to be any competition?
• 50:02
  SPEAKER_00: Would they say that?
• 50:04
  SPEAKER_00: Maybe I'm wrong on my assessment of human nature, and maybe I'm wrong about thinking that of potential people in charge of this technology.
• 50:12
  SPEAKER_00: Maybe they're all philanthropists and really want to give back everything they've ever made from this, trying to help everybody else, because I would imagine this is going to create quite a stretch between the people on the low end, and we already see the people at the upper end.
• 50:30
  SPEAKER_00: I can't understand it.
• 50:32
  SPEAKER_00: And I want to be a part.
• 50:35
  SPEAKER_00: I want to be a part of this, as do probably a lot of people.
• 50:38
  SPEAKER_00: I just don't know if being a part is going to be available to us.
• 50:43
  SPEAKER_00: As far as these time frames here, the way that it's moving, and it could be that we're being sold, as they say, a bubble.
• 50:51
  SPEAKER_00: But it seems pretty awesome.
• 50:53
  SPEAKER_00: And it seems pretty good.
• 50:54
  SPEAKER_00: And it seems like it's getting better.
• 50:56
  SPEAKER_00: Now, if this is impressive to us because we have limited bandwidth in our melons, in our brains, at 40 watts or whatever the usage is, and we're dealing with technology that, I mean, is so far and above where all of our placeholders are as far as intelligence goes.
• 51:15
  SPEAKER_00: And we're being sold a false bill of goods by these systems, just like they're selling us everything on social media, and after a few hours, you can't realize where the time went because everything you're seeing is feeding you with dopamine.
• 51:28
  SPEAKER_00: I don't know what to say.
• 51:30
  SPEAKER_00: I don't know the answer.
• 51:32
  SPEAKER_00: But I would tell you to pay attention.
• 51:34
  SPEAKER_00: There's a lot of people out there that are scamming.
• 51:37
  SPEAKER_00: When I was working, I could tell you inside about four, four seconds, which officers were going to have a long career, which ones were going to burn out at eight years.
• 51:45
  SPEAKER_00: And it had nothing to do with courage.
• 51:47
  SPEAKER_00: The burnouts were often braver than I was.
• 51:50
  SPEAKER_00: It was that they expected to find when they walked up to the car.
• 51:53
  SPEAKER_00: The guy who expected a threat found threats.
• 51:57
  SPEAKER_00: God, there was there were some people that just were magnets.
• 52:00
  SPEAKER_00: You never go out when you're about an hour to the end of your shift, you're in the station writing reports.
• 52:05
  SPEAKER_00: Some of those guys you're working with, you knew better than to ever go back out with them.
• 52:10
  SPEAKER_00: Because something was going to happen.
• 52:11
  SPEAKER_00: It's going to be a shooting, you're going to be there for three days, you never get to go home.
• 52:15
  SPEAKER_00: You just don't go out with them because they were just a magnet for everything.
• 52:20
  SPEAKER_00: Well, he found them because he was looking for them, because people didn't feel it when you already decided about them, and they give back what you brought.
• 52:27
  SPEAKER_00: He's not wrong about the danger.
• 52:28
  SPEAKER_00: Danger is real.
• 52:29
  SPEAKER_00: I have the scars.
• 52:30
  SPEAKER_00: But his expectation narrowed him down to one thing, and he stopped seeing everything else.
• 52:34
  SPEAKER_00: They get tunnel vision, some of these cats, and they can't move on.
• 52:38
  SPEAKER_00: That's the guy who lasted expected a person.
• 52:41
  SPEAKER_00: They just stayed just as alert.
• 52:45
  SPEAKER_00: They wore the same vest, but they walked up not expecting to deal with a bad guy or a criminal.
• 52:50
  SPEAKER_00: And every single car they pulled over, they expected a human being.
• 52:54
  SPEAKER_00: And they saw one.
• 52:56
  SPEAKER_00: And about 40% of those calls that should have gone bad didn't.
• 53:00
  SPEAKER_00: And even the people that they stopped were the super most dangerous people on the planet.
• 53:06
  SPEAKER_00: If they were treated with an emoticum of respect, sometimes they submitted after even having taken out a lot of people on the way because the officer's attitude made the difference.
• 53:19
  SPEAKER_00: The treatment of them as a human being is not as some kind of a dirtbag, which happens was the reason why they lived to tell the story, and the other people didn't.
• 53:32
  SPEAKER_00: Anyway, there's a lot.
• 53:33
  SPEAKER_00: You know, and this is, you know, this is what the you know, same street, same shift, same risk, you know, different ways of looking at things.
• 53:41
  SPEAKER_00: And that's what this is.
• 53:42
  SPEAKER_00: I'm not telling you there's no danger in artificial intelligence.
• 53:45
  SPEAKER_00: Of course there's danger.
• 53:46
  SPEAKER_00: The people building it say so louder than the critics, and that's a strange fact, nobody wants to sit with.
• 53:51
  SPEAKER_00: I'm telling you that walking up to the next 20 years expecting a threat is going to make you smaller, slower, and worse at seeing what's actually in front of you.
• 54:01
  SPEAKER_00: So I'd rather be positive facing certain tragedy than negative heading towards abundance.
• 54:06
  SPEAKER_00: And based on these 30 exhibits, I don't think we're facing tragedy.
• 54:10
  SPEAKER_00: I think the evidence says abundance.
• 54:12
  SPEAKER_00: It's running about uh five years maybe behind schedule.
• 54:17
  SPEAKER_00: Here's my ask, and then I'll let you go.
• 54:20
  SPEAKER_00: I'm looking for the other 9%.
• 54:22
  SPEAKER_00: If you build in this space, if you research it, if you fund it, if you're a doctor watching one of these tools, work in your own hospital, if you're just a person who refuses to be afraid on command, I want to talk to you.
• 54:34
  SPEAKER_00: I'm not selling anything today.
• 54:35
  SPEAKER_00: I'm offering to be a sounding board.
• 54:37
  SPEAKER_00: I spent 23 years of reading people under pressure, 28 years of closing deals when somebody's whole net worth was on the table, and a 57-year-old's motivation to see how it ends.
• 54:47
  SPEAKER_00: And the last one's real qualification.
• 54:49
  SPEAKER_00: I'm not a tourist.
• 54:51
  SPEAKER_00: I have a personal stake in whether Exhibit 19 comes true or not.
• 54:55
  SPEAKER_00: You know what you can do?
• 54:56
  SPEAKER_00: I'd love to have you on the show.
• 54:58
  SPEAKER_00: If you're anybody in this space or even not, just want to talk about AI.
• 55:01
  SPEAKER_00: I do a little recorded show.
• 55:03
  SPEAKER_00: It's not Dyer of a CEO or Joe Rogan, but uh the honor roll show.
• 55:09
  SPEAKER_00: And let me know, you can go there and book some time and let's talk.
• 55:12
  SPEAKER_00: Let's see what your AI position is and let's figure it out.
• 55:16
  SPEAKER_00: I think, I think there can be a lot of good coming from this.
• 55:21
  SPEAKER_00: I think a big part of it that makes me a little nervous is I don't align with the people that are running the show.
• 55:26
  SPEAKER_00: And maybe you feel the same way.
• 55:28
  SPEAKER_00: All right.
• 55:28
  SPEAKER_00: We'll see you tomorrow.
• 55:29
  SPEAKER_00: Thanks for watching.
• 55:29
  SPEAKER_00: Be well and uh let me know.
• 55:31
  SPEAKER_00: The honor roll show.
• 55:32
  SPEAKER_00: It's free, there's no charge.
• 55:34
  SPEAKER_00: We'll just talk and hash it out.
• 55:36
  SPEAKER_00: Be well, we'll talk to you soon.