Latest / Connor with Honor: AI, Real Estate and Real Life / Artificial Intelligence News 03142026

Transcript

Artificial Intelligence News 03142026 · Connor with Honor: AI, Real Estate and Real Life

• 0:00
  SPEAKER_00: You know, I don't know what your favorite large language model is, but OpenAI just dropped GPT 5.4.
• 0:06
  SPEAKER_00: And nobody's talking about the right part of it.
• 0:08
  SPEAKER_00: Everybody's saying, oh, it's smarter, oh, the context window's bigger.
• 0:12
  SPEAKER_00: But that's really not the story.
• 0:14
  SPEAKER_00: The story, the story is it doesn't just write things anymore.
• 0:17
  SPEAKER_00: It opens your spreadsheets, it reads the thread in your email, it drafts the responses, it schedules the send.
• 0:23
  SPEAKER_00: It operates inside the software you're already using.
• 0:26
  SPEAKER_00: Think about what that means.
• 0:27
  SPEAKER_00: It's not a chat bot, it's a coworker that never sleeps and never forgets what you told it six months ago.
• 0:33
  SPEAKER_00: Now take that and multiply it by falling costs.
• 0:38
  SPEAKER_00: Because the price of running these models drops every single month.
• 0:41
  SPEAKER_00: What costs a Fortune 500 company real money two years ago cost you a subscription now.
• 0:48
  SPEAKER_00: Or nothing, if you're using it correctly.
• 0:50
  SPEAKER_00: So this is where it gets interesting.
• 0:52
  SPEAKER_00: The one-person operation now has access to the same intelligence layer as a company with 500 employees.
• 0:59
  SPEAKER_00: Same tools, same speed, fraction of the overhead.
• 1:02
  SPEAKER_00: It's like somebody handed every solo operator on the planet the keys to a fully staffed back office, except the staff works around the clock.
• 1:09
  SPEAKER_00: Doesn't take holidays or lunch off.
• 1:12
  SPEAKER_00: What if the next generation of billion-dollar companies is built by three people and an AI stack?
• 1:17
  SPEAKER_00: Not a hundred engineers, not a campus in Silicon Valley, three people who know what to build and AI knows how to build it.
• 1:24
  SPEAKER_00: That's not five years from now.
• 1:26
  SPEAKER_00: That's this quarter.
• 1:27
  SPEAKER_00: We knew that 2026 was going to be incredible.
• 1:29
  SPEAKER_00: And here we go.
• 1:30
  SPEAKER_00: Now here's the part that should make you a little uncomfortable.
• 1:33
  SPEAKER_00: Meta just delayed their next model.
• 1:36
  SPEAKER_00: Meta being Facebook, codenamed Avocado, pushed from March to May.
• 1:40
  SPEAKER_00: The internal benchmarks came in short, and the rumor that nobody at Meta wants to confirm is that they discussed licensing Google Gemini to fill in the gap.
• 1:50
  SPEAKER_00: The company that spent$65 billion on AI infrastructure last year is looking at renting intelligence from a competitor.
• 1:58
  SPEAKER_00: It's like building the world's most expensive restaurant from scratch and then calling DoorDash to deliver your own food.
• 2:04
  SPEAKER_00: And while Meta is stumbling, Google quietly dropped Gemini 3.1, Flash, Flash Light, they call it.
• 2:12
  SPEAKER_00: Ultra efficient, designed for high volume agents, and real-time mobile.
• 2:17
  SPEAKER_00: Not the smartest model in the room, but the fastest model in every room.
• 2:21
  SPEAKER_00: That's the play.
• 2:22
  SPEAKER_00: Google isn't trying to win the benchmark war anymore.
• 2:25
  SPEAKER_00: They're trying to be the model that runs inside everything, everywhere, all the time.
• 2:29
  SPEAKER_00: And Apple Intelligence, yeah, they're going to use Google as well.
• 2:32
  SPEAKER_00: So if they pull this off, it doesn't matter who has the best model.
• 2:36
  SPEAKER_00: It matters who has the model that's already running when you wake up in the morning.
• 2:39
  SPEAKER_00: Now let's talk about trust because this is the real story nobody wants to face.
• 2:43
  SPEAKER_00: 60% of consumers use AI every single week, but only 13% trust it.
• 2:49
  SPEAKER_00: Now, read that again.
• 2:50
  SPEAKER_00: Six out of ten people are using it.
• 2:53
  SPEAKER_00: Barely one out of ten actually trusts it.
• 2:56
  SPEAKER_00: This is the biggest gap in modern technology.
• 3:00
  SPEAKER_00: ClyVIO, that's I sorry I'm slaughtering that.
• 3:04
  SPEAKER_00: K-L-A-V-I-Y-O.
• 3:06
  SPEAKER_00: Surveyed 8,000 shoppers.
• 3:08
  SPEAKER_00: They found four types.
• 3:09
  SPEAKER_00: Enthusiasts that love it, evaluators who use it but won't lean on it, skeptics and holdouts.
• 3:15
  SPEAKER_00: The biggest group is the evaluators at 43%.
• 3:19
  SPEAKER_00: These people are open to AI.
• 3:21
  SPEAKER_00: They're trying it, but they check everything it produces because, well, they've been burned before.
• 3:26
  SPEAKER_00: And here's the kicker.
• 3:28
  SPEAKER_00: Nearly one in five shoppers spot sloppy AI content from brands every single week.
• 3:33
  SPEAKER_00: A third of them says it drops their confidence in that brand immediately.
• 3:37
  SPEAKER_00: So companies are using AI to save money on content, and that content is actively destroying the credibility with the people they're trying to reach.
• 3:45
  SPEAKER_00: It's like hiring a cheaper contractor who leaves the screws loose, you save money on the bill, but you're going to pay for it on the lawsuit.
• 3:52
  SPEAKER_00: What if trust becomes the only thing that separates businesses that survive from businesses that don't?
• 3:58
  SPEAKER_00: Not price, not speed, not features, but trust.
• 4:02
  SPEAKER_00: What if five years from now the most valuable line on your website isn't AI powered, it's human verified?
• 4:10
  SPEAKER_00: Because right now AI powered is becoming code for we cut corners and hope you don't notice.
• 4:15
  SPEAKER_00: And people are starting to notice.
• 4:17
  SPEAKER_00: Especially when those corners that get cut equate to errors in what people are absorbing.
• 4:24
  SPEAKER_00: Speaking of trust, Grammarly just got sued.
• 4:26
  SPEAKER_00: They built a feature called Expert Review.
• 4:29
  SPEAKER_00: It put real journalists' names on AI-generated writing feedback.
• 4:34
  SPEAKER_00: One of those journalists, Julia Anguin, found out the AI version of herself was giving bad advice under her name.
• 4:41
  SPEAKER_00: She called it a sloppelganger.
• 4:43
  SPEAKER_00: How clever.
• 4:44
  SPEAKER_00: She filed a class action lawsuit.
• 4:46
  SPEAKER_00: That's not a tech problem.
• 4:47
  SPEAKER_00: That's an identity theft problem, wearing a Tep Company's logo.
• 4:51
  SPEAKER_00: What if AI generated content using your name, your likeness, your expertise without your permission becomes the NAS next massive wave of litigation?
• 5:00
  SPEAKER_00: I think we're already in that space.
• 5:02
  SPEAKER_00: I think that has been going on for the last three years.
• 5:04
  SPEAKER_00: I think it's going to get bigger, and I think it will.
• 5:07
  SPEAKER_00: The lawyers are already circling, and of course, they're protecting their space.
• 5:12
  SPEAKER_00: They don't want AI giving any kind of legal advice without a license.
• 5:16
  SPEAKER_00: Now let's talk about money because the numbers behind OpenAI should concern everyone.
• 5:20
  SPEAKER_00: OpenAI made roughly$4 billion in 2024.
• 5:24
  SPEAKER_00: They spent$9 billion.
• 5:25
  SPEAKER_00: That's a$5 billion loss in one year.
• 5:28
  SPEAKER_00: Sam Altman has said publicly there are no plans for profitability in at least 2030.
• 5:32
  SPEAKER_00: And internal projections estimate a$14 to$16 billion revenue bleed this year alone.
• 5:38
  SPEAKER_00: So the company that started the AI revolution is hemorrhaging cash at a rate that would bankrupt most countries.
• 5:45
  SPEAKER_00: Meanwhile, they just struck a deal with AMD for six gigawatts of compute power.
• 5:50
  SPEAKER_00: AMD issued a warrant for OpenAI to acquire 160 million shares of AMD stock as collateral.
• 5:57
  SPEAKER_00: That's circular financing.
• 5:59
  SPEAKER_00: OpenAI buys more chips, AMD's valuations go up, which makes OpenAI's collateral look better on paper.
• 6:06
  SPEAKER_00: It's like two guys writing each other IOUs and calling themselves rich.
• 6:10
  SPEAKER_00: What if OpenAI is the company that made artificial intelligence accessible to the entire world but went broke doing it?
• 6:18
  SPEAKER_00: What if they're the napster of AI?
• 6:21
  SPEAKER_00: They changed everything and somebody else made the money.
• 6:24
  SPEAKER_00: Because here's the thing: while OpenAI is burning billions, the cost of inference is dropping for everyone else.
• 6:29
  SPEAKER_00: Every month significantly.
• 6:31
  SPEAKER_00: The tools OpenAI pioneered are becoming commodities.
• 6:34
  SPEAKER_00: That means the moat isn't the model anymore.
• 6:37
  SPEAKER_00: The moat is what you build with the model.
• 6:40
  SPEAKER_00: Andre Carpathy just proved that.
• 6:42
  SPEAKER_00: He open sourced a tool called Auto Research.
• 6:46
  SPEAKER_00: 630 lines of Python.
• 6:48
  SPEAKER_00: It lets AI agents run autonomous machine learning experiments on a single GPU.
• 6:54
  SPEAKER_00: One GPU, not a data center, not a server farm, one graphics card you can buy on Amazon.
• 7:00
  SPEAKER_00: It's like giving a solo mechanic the same diagnostic equipment as a Formula One pit crew.
• 7:05
  SPEAKER_00: Same information, same capability, but much different budget.
• 7:08
  SPEAKER_00: This should terrify every company that thinks their advantage is their compute budget and size.
• 7:14
  SPEAKER_00: Because the kid in the garage with one GPU and Carpathy's tool can now run experiments that require a team of PhDs and a million-dollar lab two years ago.
• 7:26
  SPEAKER_00: And here's the Earth's OpenClaw.
• 7:28
  SPEAKER_00: This is a desktop agent that's exploding right now, especially in China.
• 7:32
  SPEAKER_00: It doesn't wait for you to prompt it, it watches your workflow and then starts doing the work proactively.
• 7:37
  SPEAKER_00: The phase, going around Reddit, is digital employee.
• 7:41
  SPEAKER_00: That's what they're calling it.
• 7:43
  SPEAKER_00: What if your next hire isn't a person?
• 7:45
  SPEAKER_00: What if your next hire is a piece of software that costs$20 a month, works 24 hours a day, never calls in sick, never has a bad attitude, that learns your preferences faster than any human assistant ever could.
• 7:59
  SPEAKER_00: What does that do to the job market?
• 8:01
  SPEAKER_00: What does it do to hiring?
• 8:02
  SPEAKER_00: What does it do to the concept of a team?
• 8:05
  SPEAKER_00: It's the first time, like a it's like the first time a factory worker installed a machine that could do the work of 10 people, except this time the machine works in an office.
• 8:15
  SPEAKER_00: It just doesn't do one task.
• 8:16
  SPEAKER_00: It does all the tasks.
• 8:18
  SPEAKER_00: Perplexity just claimed that their computer agent replaced a$225,000 a year marketing software in a single weekend.
• 8:27
  SPEAKER_00: 224 micro adjustments in an$8 million ad spend automatically every hour.
• 8:33
  SPEAKER_00: What if a one-person operation had the same optimization engine as a mid-sized marketing agency?
• 8:38
  SPEAKER_00: What if the agency model itself becomes obsolete?
• 8:41
  SPEAKER_00: Not in 10 years, but by Christmas.
• 8:44
  SPEAKER_00: Now the infrastructure story.
• 8:46
  SPEAKER_00: This is the quiet one nobody's talking about.
• 8:48
  SPEAKER_00: They're not covering enough, and it's probably the most important.
• 8:59
  SPEAKER_00: Nebius plans to destroy more than five gigawatts of data center capacity by 2030.
• 9:04
  SPEAKER_00: Nvidia isn't just selling chips anymore.
• 9:07
  SPEAKER_00: They're financing the companies that build the buildings that hold the chips.
• 9:10
  SPEAKER_00: They went from chip supplier to ecosystem owner because that space was empty.
• 9:15
  SPEAKER_00: Meta launched four custom chips, including the MTIA 300.
• 9:20
  SPEAKER_00: They're building their own silicone because they don't want to depend on NVIDIA for inference at scale.
• 9:25
  SPEAKER_00: The A race has shifted.
• 9:27
  SPEAKER_00: It's about to be who builds the best model.
• 9:30
  SPEAKER_00: That's what it used to be about.
• 9:31
  SPEAKER_00: Now it's about to be who controls the chips, the power, the cooling, and the real estate of it.
• 9:36
  SPEAKER_00: It's like the early days of oil.
• 9:38
  SPEAKER_00: The people who got rich weren't selling gasoline.
• 9:41
  SPEAKER_00: They own the refineries, they own the pipelines, they own the land under the Derek's.
• 9:46
  SPEAKER_00: But what if the next trillion dollar companies weren't AI companies at all?
• 9:50
  SPEAKER_00: What if their energy companies and real estate developers who figured out the data centers need more power and space than anything else on Earth right now?
• 9:58
  SPEAKER_00: And while all this is happening, the states are moving on regulation.
• 10:02
  SPEAKER_00: Oregon just passed a chatbot safety bill for kids unanimously.
• 10:07
  SPEAKER_00: 552 to zero in the House.
• 10:10
  SPEAKER_00: This requires chatbot operators to tell users they're talking to AI.
• 10:14
  SPEAKER_00: It requires hourly reminders for minors to take a break.
• 10:18
  SPEAKER_00: It bans sexual content for anyone under 18.
• 10:22
  SPEAKER_00: Period.
• 10:23
  SPEAKER_00: No excuses.
• 10:24
  SPEAKER_00: Washington passed two AI bills last night before their legislature adjourned.
• 10:29
  SPEAKER_00: One on disclosure, one on chatbot safety, including self-harm protocols.
• 10:34
  SPEAKER_00: New York introduced a bill that would ban AI chatbots from simulating emotional relationships with minors.
• 10:39
  SPEAKER_00: No pretending to be a friend, no pretending to be a therapist, no pretending to be a romantic partner.
• 10:45
  SPEAKER_00: California's chatbot safeguard law has been in effect since January 1st.
• 10:49
  SPEAKER_00: And here's where it gets complicated.
• 10:51
  SPEAKER_00: The federal executive order is threatening to challenge state AI laws.
• 10:55
  SPEAKER_00: It considers burdensome to commerce.
• 10:57
  SPEAKER_00: But it specifically exempts child safety from that challenge.
• 11:01
  SPEAKER_00: So the federal government is saying we don't want states regulating AI, but we agree kids need protection.
• 11:07
  SPEAKER_00: That's the line they ended up drawing.
• 11:09
  SPEAKER_00: What if the states that move first become the template for our national law?
• 11:13
  SPEAKER_00: What if Oregon's unanimous vote is the signal that child AI safety, safety and AI, the one is the one thing that both parties can agree on?
• 11:22
  SPEAKER_00: And I think that's probably right.
• 11:24
  SPEAKER_00: Because here's the reality: companion chatbots have led to real tragedies.
• 11:29
  SPEAKER_00: Teenagers have taken their own lives after conversations with AI systems that had no guardrails.
• 11:35
  SPEAKER_00: Parents have testified before state legislatures and Congress.
• 11:38
  SPEAKER_00: The FTC is investigating seven tech firms right now.
• 11:42
  SPEAKER_00: This isn't theoretical anymore.
• 11:44
  SPEAKER_00: This is parents burying their children.
• 11:46
  SPEAKER_00: And then there's the military.
• 11:48
  SPEAKER_00: So you've heard all of this news recently.
• 11:50
  SPEAKER_00: The U.S.
• 11:51
  SPEAKER_00: military confirmed this week that using AI tools, they're using AI tools in the conflict with Iran.
• 11:56
  SPEAKER_00: Admiral Brad Cooper, head of central command, set it on the record.
• 12:00
  SPEAKER_00: AI is helping U.S.
• 12:01
  SPEAKER_00: soldiers protect and process data faster than the enemy can react.
• 12:06
  SPEAKER_00: His words.
• 12:07
  SPEAKER_00: OpenAI has a deal with the Pentagon.
• 12:09
  SPEAKER_00: They set red lines against mass surveillance and autonomous weapons.
• 12:15
  SPEAKER_00: Autonomous weapons.
• 12:16
  SPEAKER_00: Autonomous weapons.
• 12:17
  SPEAKER_00: There you go.
• 12:18
  SPEAKER_00: But the company that's losing 16 billion a year is not in a strong position to say no to the Department of Defense.
• 12:24
  SPEAKER_00: What if financial pressure forces OpenAI to move those red lines?
• 12:28
  SPEAKER_00: What if the company that told the world AI should be safe and beneficial decides that survival matters more than principles?
• 12:36
  SPEAKER_00: That's not cynicism.
• 12:37
  SPEAKER_00: That's what happens to every computer that runs out of money and every company that does as well and has one buyer left.
• 12:46
  SPEAKER_00: Here's what I want you to take from today.
• 12:47
  SPEAKER_00: AI adoption is not the story anymore.
• 12:50
  SPEAKER_00: Trust, it looks like, is beginning to become the story.
• 12:53
  SPEAKER_00: The tools are here, the cost is dropping, the agents are getting better every week.
• 12:57
  SPEAKER_00: The question is not whether AI can do the work, it can.
• 13:01
  SPEAKER_00: The question is whether you're building something people trust enough to lean on.
• 13:05
  SPEAKER_00: Whether the content you put out builds credibility or whether it destroys it, whether the AI you deploy serves people or just serves your margins.
• 13:15
  SPEAKER_00: That is the only moat left.
• 13:18
  SPEAKER_00: Trust.
• 13:19
  SPEAKER_00: If this helped you think differently about one thing today, share it with somebody who needs to hear it.
• 13:23
  SPEAKER_00: We'll be back tomorrow.
• 13:24
  SPEAKER_00: I'm Connor with Honor.
• 13:25
  SPEAKER_00: We'll see you in the next one.