Latest / Connor with Honor: AI, Real Estate and Real Life / AI Productivity Sounds Great Until We Measure the Result

Transcript

AI Productivity Sounds Great Until We Measure the Result · Connor with Honor: AI, Real Estate and Real Life

• 0:00
  SPEAKER_00: Artificial intelligence just got his own jalapeno.
• 0:03
  SPEAKER_00: And that's not a metaphor.
• 0:05
  SPEAKER_00: OpenAI built a computer chip and named it jalapeno.
• 0:08
  SPEAKER_00: Company says the chip can serve artificial intelligence faster while using less power, maybe more spicy.
• 0:15
  SPEAKER_00: Well, on that same day, OpenAI released an administrative tool that can watch company usage, change permissions, manage spending limits, and handle routine requests through a conversation.
• 0:25
  SPEAKER_00: Then the United States government published research saying businesses often cannot clearly connect their artificial intelligence plans to the economic results they expected.
• 0:36
  SPEAKER_00: We bought in a little too quick, apparently.
• 0:38
  SPEAKER_00: The machine is getting faster.
• 0:39
  SPEAKER_00: Management is getting easier.
• 0:41
  SPEAKER_00: Proof is still wandering around the parking lot trying to find the entrance.
• 0:45
  SPEAKER_00: And that's our show today.
• 0:46
  SPEAKER_00: Three developments for the last 24 hours.
• 0:48
  SPEAKER_00: One custom chip, one administrative tool, and one government report carrying a bucket of cold water, and one question running through all three.
• 0:56
  SPEAKER_00: Are we creating more value or are we getting better at describing the value we hope to create?
• 1:01
  SPEAKER_00: OpenAI published that first measured results for jalapeno, custom chip built specifically to run artificial intelligence models.
• 1:10
  SPEAKER_00: Inference is the part where a trained model actually answers us.
• 1:13
  SPEAKER_00: Training is the school.
• 1:14
  SPEAKER_00: Inference is the shift after graduation when somebody finally expects the kid to earn rent.
• 1:20
  SPEAKER_00: OpenAI says jalapeno completed between 1.5 and 1.9 times more artificial intelligence work per watt at peak throughput than the commercial systems and its comparison.
• 1:32
  SPEAKER_00: It also reported between 1.7 and 3.6 times lower to end-to-end delay across three public models.
• 1:38
  SPEAKER_00: That means more answers from the same amount of electricity with less weighting.
• 1:41
  SPEAKER_00: The company tested the chip with a public benchmark called Inference X.
• 1:46
  SPEAKER_00: A benchmark is a standardized test for machines.
• 1:48
  SPEAKER_00: It's the artificial intelligence version of lining up 10 pickup trucks, loading them with concrete, and seeing which one gets over the hill without leaving a transmission laying on the asphalt.
• 1:57
  SPEAKER_00: And here's the catch.
• 1:58
  SPEAKER_00: OpenAI built the truck, picked the setup, and of course ran the test and published the brochure.
• 2:05
  SPEAKER_00: The benchmark's public, the reported numbers are specific.
• 2:08
  SPEAKER_00: That's better than a stage presentation full of glowing circles and the word revolutionary spelled incorrectly.
• 2:15
  SPEAKER_00: But these are still company results.
• 2:16
  SPEAKER_00: They're not yet an independent verdict from years of production use.
• 2:21
  SPEAKER_00: OpenAI says it plans to begin deploying jalapeno inside its own computer systems by the end of the year.
• 2:27
  SPEAKER_00: Plans are not deployments.
• 2:29
  SPEAKER_00: The chip is real enough to measure, but the promised economic effect is still ahead of us.
• 2:34
  SPEAKER_00: Why does this matter to a family, a worker, a small business owner who's never spent a morning comparing token latency?
• 2:41
  SPEAKER_00: Well, because that cost of running artificial intelligence decides where it can go.
• 2:46
  SPEAKER_00: When each answer becomes cheaper and faster, artificial intelligence stops being something we visit for a clever paragraph, but it starts becoming something that stays on the job.
• 2:58
  SPEAKER_00: It can answer every phone call, review every incoming document, watch every service request, now check every invoice, follow every customer conversation.
• 3:06
  SPEAKER_00: Not because it suddenly developed a work ethic, but because somebody removed the meter that made the constraint attention too expensive.
• 3:15
  SPEAKER_00: Well, the name jalapeno is almost too perfect.
• 3:18
  SPEAKER_00: The chip makes the system hotter, faster, and cheaper.
• 3:21
  SPEAKER_00: Somewhere a marketing department looked at a 700-watt processor and said, what this needs is the name of something we warn our children not to rub into their eyes.
• 3:30
  SPEAKER_00: The larger move is not the name.
• 3:32
  SPEAKER_00: OpenAI is trying to control more of the chain.
• 3:34
  SPEAKER_00: The models, the software, the chip, the memory, the network, the data centers, the products we use.
• 3:39
  SPEAKER_00: That's called vertical integration.
• 3:42
  SPEAKER_00: Plain English, they're trying to own a lot more of the restaurant, not just the recipe.
• 3:47
  SPEAKER_00: The stove, the delivery truck, the building, and eventually the person and person asking.
• 3:52
  SPEAKER_00: You want fries with that.
• 3:54
  SPEAKER_00: But you can lower costs.
• 3:55
  SPEAKER_00: It can concentrate power.
• 3:56
  SPEAKER_00: The same company that builds the intelligence could increasingly control the machinery that delivers it.
• 4:02
  SPEAKER_00: For regular people, the opportunity is real.
• 4:04
  SPEAKER_00: Cheaper intelligence means smaller businesses can use tools only large companies could afford a few years ago.
• 4:10
  SPEAKER_00: The guardrail is also real.
• 4:12
  SPEAKER_00: We should not build a business that becomes helpless when one provider changes a price, removes a model, or tightens access.
• 4:19
  SPEAKER_00: Use the best tool, keep our data portable, keep your instructions documented.
• 4:24
  SPEAKER_00: Know what the system does when the preferred model is unavailable.
• 4:28
  SPEAKER_00: A spare tire is boring right up until the freeway shoulder.
• 4:32
  SPEAKER_00: The same development moves artificial intelligence deeper into management.
• 4:35
  SPEAKER_00: OpenAI introduced an admin plugin for Jet GPT work and codecs.
• 4:41
  SPEAKER_00: An administrator can ask questions about workspace activity, review credit usage, add or remove members, change groups, inspect permissions, adjust usage limits, and handle supported requests through a conversation.
• 4:53
  SPEAKER_00: That tool follows the administrator's existing permissions.
• 4:57
  SPEAKER_00: It does not magically give somebody access that they didn't already have.
• 5:01
  SPEAKER_00: And that does matter.
• 5:03
  SPEAKER_00: The dangerous version of Artificial Intelligence Administrator is not a robot carrying a tie or wearing a tie and carrying a clipboard.
• 5:10
  SPEAKER_00: It's a system with broad permissions, a vague instruction, and nobody checking what happened after it said it was finished.
• 5:16
  SPEAKER_00: OpenAI says its own information technology team uses an agent to handle employee requests.
• 5:21
  SPEAKER_00: The company reports that developed workflows resolved about 45% of ticket volume.
• 5:26
  SPEAKER_00: It also says support volume roughly doubled while the backlog was eliminated.
• 5:31
  SPEAKER_00: These are company reported results from the company that's selling the tool.
• 5:35
  SPEAKER_00: Useful evidence, but not holy scripture.
• 5:37
  SPEAKER_00: Still, 45% is large enough to matter.
• 5:40
  SPEAKER_00: It nearly half the retween tickets can be resolved automatically.
• 5:43
  SPEAKER_00: The human job starts to change.
• 5:45
  SPEAKER_00: The person is no longer spending the entire day resetting access or finding a policy and explaining it for the 83rd time.
• 5:54
  SPEAKER_00: That turning the laptop off and back on is not a spiritual philosophy.
• 5:58
  SPEAKER_00: The human can work on exceptions, security judgment.
• 6:02
  SPEAKER_00: The strange cases where the request sounds normal until we notice the employee left the company six months ago and is emailing from a fishing boat near Belize.
• 6:11
  SPEAKER_00: That's the useful version of artificial deployment.
• 6:14
  SPEAKER_00: Routine work goes to the machine.
• 6:16
  SPEAKER_00: Authority stays bounded.
• 6:18
  SPEAKER_00: Exceptions go to a human.
• 6:19
  SPEAKER_00: Every action leaves a record, but we need to pressure test the labor story.
• 6:23
  SPEAKER_00: If one employee becomes three times more productive, the company doesn't automatically hire three times as many employees.
• 6:29
  SPEAKER_00: That can also happen if demand can absorb three times the output.
• 6:33
  SPEAKER_00: If the company's a fixed amount of work management may reduce headcount and keep the productivity gain, that's not pessimism.
• 6:41
  SPEAKER_00: That's arithmetic wearing work boots.
• 6:43
  SPEAKER_00: Productivity doesn't have to become unemployment.
• 6:45
  SPEAKER_00: A company can use the extra capacity to improve service, shorten response times, build products it can never afford, expand into new markets, and let human beings do the judgment-heavy work machines cannot own.
• 6:57
  SPEAKER_00: But that output or outcome requires a decision.
• 7:00
  SPEAKER_00: The technology does not choose whether the saved time becomes better service, more growth, shorter work weeks, higher profit, or fewer employees.
• 7:08
  SPEAKER_00: Management chooses that.
• 7:09
  SPEAKER_00: And that leads directly to the third development.
• 7:12
  SPEAKER_00: The United States Bureau of Economic Analysis published a research spotlight examining what businesses expected from artificial intelligence and what actually happened later.
• 7:22
  SPEAKER_00: The Bureau of Economic Analysis is the governmental group that measures major parts of the economy.
• 7:27
  SPEAKER_00: It's not selling an artificial intelligence subscription.
• 7:31
  SPEAKER_00: Its incentive is better measurement, although the researchers clearly state that the available data has limits.
• 7:37
  SPEAKER_00: The study found that businesses predicted their future artificial intelligence use six months ahead with two percentage points on average.
• 7:44
  SPEAKER_00: That sounds impressive.
• 7:46
  SPEAKER_00: Companies were getting fairly good at predicting whether they would use AI or artificial intelligence.
• 7:51
  SPEAKER_00: Using it was not the same as getting the result they wanted.
• 7:54
  SPEAKER_00: The researchers compared reasons companies gave for adopting artificial intelligence with economic measures later.
• 8:01
  SPEAKER_00: Early adopters sometime achieved their goals, but often the researchers could not find a clear connection between the original motivation and the later economic result.
• 8:10
  SPEAKER_00: The data were too coarse to isolate every effect.
• 8:13
  SPEAKER_00: So this report doesn't prove artificial intelligence failed.
• 8:17
  SPEAKER_00: It proves something less dramatic and actually more useful.
• 8:20
  SPEAKER_00: Installing a tool is easy to count.
• 8:22
  SPEAKER_00: Proving the tool changed the business is harder.
• 8:25
  SPEAKER_00: We've seen this movie before.
• 8:27
  SPEAKER_00: The company buys software.
• 8:29
  SPEAKER_00: Everybody attends a webinar, new dashboard appears, three people receive a certificate.
• 8:34
  SPEAKER_00: Six months later, nobody remembers the password, but the annual renewal knows exactly where that credit card lives.
• 8:41
  SPEAKER_00: Adoption is not value.
• 8:43
  SPEAKER_00: Activity is not value.
• 8:44
  SPEAKER_00: Prompts entered are not value either.
• 8:47
  SPEAKER_00: Reports generated are not value.
• 8:49
  SPEAKER_00: Value is the customer reached faster.
• 8:51
  SPEAKER_00: The mistake caught before it left the building, the invoice collected sooner.
• 8:56
  SPEAKER_00: The product shipped that could not be built before.
• 8:59
  SPEAKER_00: The hour returned to a human being.
• 9:00
  SPEAKER_00: The deep question today is whether cheaper intelligence automatically creates more economic activity.
• 9:06
  SPEAKER_00: OpenAI argues that efficiency can make more uses affordable.
• 9:11
  SPEAKER_00: Economists call this the Javon's paradox.
• 9:14
  SPEAKER_00: When a resource becomes more efficient, total use can rise because new uses become practical.
• 9:21
  SPEAKER_00: That can happen with artificial intelligence.
• 9:23
  SPEAKER_00: Cheaper answers can create services that did not exist with every task, required an expensive professional.
• 9:30
  SPEAKER_00: A small business could review every consumer call instead of sampling five.
• 9:35
  SPEAKER_00: A family could compare confusing medical bills before making calls.
• 9:39
  SPEAKER_00: A worker could rehearse a difficult conversation with the words stop getting stuck behind the ribs.
• 9:45
  SPEAKER_00: Well, cheaper supply doesn't mean manufacture demand by itself.
• 9:50
  SPEAKER_00: If a bakery can produce three times as many cakes, but the town still eats the same number of cakes, we don't have an ambulance or an abundance miracle.
• 9:59
  SPEAKER_00: We have a freezer problem.
• 10:00
  SPEAKER_00: The opportunity is to use extra capacity for work that was previously ignored.
• 10:06
  SPEAKER_00: Faster follow-up, better documentation, new offers, more personal service, deeper checking.
• 10:11
  SPEAKER_00: That's how productivity can become growth instead of a headcount meeting with a suspiciously cheerful calendar title.
• 10:18
  SPEAKER_00: Now there's another mistake hiding in the productivity story.
• 10:21
  SPEAKER_00: We keep measuring how fast the machine produces the first draft.
• 10:25
  SPEAKER_00: We do not always measure how long the human spends checking it.
• 10:29
  SPEAKER_00: If artificial intelligence writes a proposal in four minutes and a person spends two hours finding the invented numbers, we didn't save two hours.
• 10:38
  SPEAKER_00: We bought a very fast liar and assigned somebody to follow it around with a mop.
• 10:44
  SPEAKER_00: That doesn't make the tool useless.
• 10:46
  SPEAKER_00: It means the quality control belongs inside the measurement.
• 10:50
  SPEAKER_00: The same applies at home.
• 10:51
  SPEAKER_00: Artificial intelligence can compare insurance plans, summarize a school policy, organize a family budget, or explain a medical bill.
• 10:59
  SPEAKER_00: That can save real time, but the final decision still touches our money, our children, and our health.
• 11:05
  SPEAKER_00: The tool can prepare the table, but it doesn't get to decide what the family eats.
• 11:09
  SPEAKER_00: This is where regular people can outperform companies with larger budgets.
• 11:13
  SPEAKER_00: A small operation can see the entire workflow.
• 11:15
  SPEAKER_00: We know who called.
• 11:17
  SPEAKER_00: We know what went wrong.
• 11:19
  SPEAKER_00: We know whether the consumer came back.
• 11:23
  SPEAKER_00: Large companies can buy a failed artificial intelligence product under 11 departments, four dashboards and a meeting called strategic enablement alignment.
• 11:32
  SPEAKER_00: But the time or by the time everybody agrees it didn't work.
• 11:35
  SPEAKER_00: The software has a parking space and dental coverage.
• 11:39
  SPEAKER_00: We can move faster because we can measure closer to the ground.
• 11:42
  SPEAKER_00: Did the phone get answered?
• 11:43
  SPEAKER_00: Did the consumer get served?
• 11:45
  SPEAKER_00: Did the error rate fall?
• 11:47
  SPEAKER_00: Did we recover time?
• 11:48
  SPEAKER_00: Did revenue move?
• 11:49
  SPEAKER_00: Those questions aren't glamorous, but neither is checking the oil.
• 11:53
  SPEAKER_00: But the engine prefers it to a motivational speech.
• 11:56
  SPEAKER_00: Here's the practical development example.
• 11:58
  SPEAKER_00: Take one repeated task in a business.
• 12:00
  SPEAKER_00: Customer calls are a good example.
• 12:03
  SPEAKER_00: Do not begin with announcing that artificial intelligence will transform customer experience.
• 12:08
  SPEAKER_00: That sentence has escaped from too many conference rooms already.
• 12:12
  SPEAKER_00: Begin with a number.
• 12:14
  SPEAKER_00: How many calls came in last week?
• 12:15
  SPEAKER_00: How many were answered?
• 12:17
  SPEAKER_00: How long did the callers wait?
• 12:19
  SPEAKER_00: How many needed a human decision?
• 12:21
  SPEAKER_00: And then give artificial intelligence one narrow job.
• 12:24
  SPEAKER_00: Answer after hours.
• 12:26
  SPEAKER_00: Collect the caller's name, reason, urgency, and preferred callback time.
• 12:30
  SPEAKER_00: Don't let it negotiate prices.
• 12:32
  SPEAKER_00: Don't let it promise outcomes.
• 12:34
  SPEAKER_00: Don't let it improve policy because it felt creative after dinner.
• 12:38
  SPEAKER_00: Run it for 30 days.
• 12:39
  SPEAKER_00: Compare missed calls, response times, appointments, complaints, and human hours before and after.
• 12:44
  SPEAKER_00: That's deployment with judgment.
• 12:46
  SPEAKER_00: Small lane, clear authority, measurable result, human escalation.
• 12:51
  SPEAKER_00: The guardrail is permissions.
• 12:54
  SPEAKER_00: An artificial intelligence system should receive the least authority required to do its job.
• 13:00
  SPEAKER_00: Read access before write access, draft before send, recommendation before approve, one consumer record before the entire database.
• 13:10
  SPEAKER_00: The speed of the machine makes permission mistakes more dangerous, not less.
• 13:14
  SPEAKER_00: A person can make one bad change before lunch.
• 13:17
  SPEAKER_00: An automated system can make 10,000 and still be early for the meeting.
• 13:22
  SPEAKER_00: Our move today is simple.
• 13:23
  SPEAKER_00: Pick one artificial intelligence workflow already running in our life or business.
• 13:27
  SPEAKER_00: Write down the result it's supposed to create.
• 13:30
  SPEAKER_00: Write down the number that proves it.
• 13:32
  SPEAKER_00: Write down the authority it actually needs.
• 13:35
  SPEAKER_00: If we cannot name the result, the measurement, the boundary, we don't have deployment.
• 13:40
  SPEAKER_00: We have a science project with access to the company credit card.
• 13:43
  SPEAKER_00: The chips will get faster, the agents will take more action, the promotional claims will arrive wearing better shoes.
• 13:49
  SPEAKER_00: Our advantage is not predicting every turn, it's the learning how to measure the distance, control the authority, and make the move that produces a real result.
• 13:59
  SPEAKER_00: AI for everyone, not just the wealthy.
• 14:02
  SPEAKER_00: I'm Connor with Honor.
• 14:03
  SPEAKER_00: We'll see you tomorrow.