TBPN

BREAKING: @JackMa spotted courtside with Jackie Chan and David Beckham at the NBA China Games

Union Square Ventures’ @rebeccakaden and @mignano explain why NYC is poised to thrive as AI shifts from an infrastructure buildout into a deployment and application era:

1. Consumer-oriented talent
2. Cross-industry density

“One of our portfolio companies, Suno, they have offices in San Francisco, they have offices in Boston, but their biggest office is actually becoming New York City.”

“It’s strategically advantageous to maybe build around the market there and attract talent that is interested in consumer and other sort of non-pure enterprise or infra areas.”

"The average consumer actually does not value their time. You would think they do, but they really do not.”

"What people do care about is saving or making money.”

@zach_yadegari, founder of Persona, explains how he's building his AI assistant around that belief:

Arena's @ml_angelopoulos says one of the hardest problems in alignment is that users often can't tell when the model is helping them or hurting them.

"You really need alignment signals that are independent. We have three at Arena."

"The first is that models take unauthorized actions, which means that you permission them in a certain way, but they break those permissions."

"It can cause things like the Hugging Face incident. But it can also cause very mundane problems like losing important information within your company or on your own laptop."

"The second signal is called deceptive completion."

"The model will tell you that it did something, but it didn't actually do it."

"Third is false attribution. It'll attribute intent to the user when that intent was not supposed to be there."

"If models are able to do this, then certainly they're not perfectly safe for a user and they're not perfectly aligned to user intent. They might actually hurt people down the line."

Arena CEO @ml_angelopoulos says using the wrong model is "so much worse than people realize."

"People will literally be tipping their model $15 just by saying thank you at the end of the day."

.@nathanbenaich explains why AI labs are shifting compute from pretraining to RL:

"The last 4-5 years have been trying to figure out what the recipe is for pretraining, what is the best data mixture, what are the best ingredients to this whole magical soup."

"Then over the years we figured out how to do that better. So it strikes me as normal that one would end up spending less money on exploratory work because the solution is more in the plain eye."

"The whole pitch with RL is this idea that what we have as solutions today are really local maxima that were discovered through human ingenuity and sharing, through reading and whatnot. But that doesn't necessarily mean it's the global maxima."

Air Street Capital GP @nathanbenaich says 99% of people still have no idea what to do with AI because those building it aren't very effective at educating people on how to use it, so the labs should take a page out of Apple's book:

"I would love there to be a Genius Bar for AI."

"You either die getting to the frontier, or you live long enough to serve inference."

Air Street Capital GP @nathanbenaich on the neolab to neocloud pivot:

"Companies that have tried to build frontier models and failed have eventually pivoted towards building huge data center businesses to serve Chinese models."

BREAKING: @hone raises $60M in seed funding co-led by Benchmark & Index Ventures

Happy Thursday. On today's show:

- @mignano (USV)
- @rebeccakaden (USV)
- @nathanbenaich (Air Street Capital)
- @ml_angelopoulos (Arena)
- @M_Sindicich (Navan)
- @healeycypher (Navan)
- @moritz_stephan (Hone)

See you on the stream.

Mathpocalypse, Cryptopocalpyse, ARRpocalypse, Housingpocalypse x.com/i/broadcasts/1…

BREAKING: @Emily_Sundberg celebrates four years of Feed Me, her newsletter about the spirit of enterprise

BREAKING: You Can See Everything has a runtime of 2h 54m, longer than The Odyssey

ChatGPT PM @AarushSelvan describes how AI has upended product management:

In the SaaS era, product management was built around mostly deterministic software.

With AI, the role is now more about defining what “good” looks like and steering a probabilistic system toward it.

“You sort of have to start from scratch a little bit and be comfortable with non-determinism.”

“Getting very comfortable with evals and how do we measure progress is something that’s a really important skill.”

“No longer can a designer build a weather card or a sports card and be like, ‘This is how it’s going to appear when it triggers,’ because the model can just generate that on the fly.”

Hudson Institute’s @wrmead says nostalgia has softened our memory of how difficult life was from the late ’60s through the early ’80s:

"We hear Zoomers saying, 'Oh, this is terrible. We’re going to be the first generation that doesn’t live as well as our parents.'"

"That's just the way I was when I was a kid!"

"The boomers were going to be the first post-American generation. We had left behind all that stupid American ideology. The war in Vietnam, the civil rights movement."

"We saw through the American red, white, and blue, blah, blah. And also, we were the least religious generation anybody had ever seen."

"We came out of college into the stagflation of the 1970s. When we started wanting to buy houses in 1980, mortgage rates were 21%. We were so screwed. That’s how we felt."

“And then, of course, things kind of worked out. Not for everybody, but for a lot of us it ended up working out okay."

“I think that makes people like me resonate emotionally with the way people are feeling their situation, but we also don’t think the past was as utopian as some might say. And yet it did work out."

.@wrmead believes AI will turn services that were once luxuries for the ultra-wealthy into everyday conveniences for ordinary people.

"You had to be super rich in the 1800s for the concept of fashion to have any connection with your life."

"A working definition of economic progress is things that only the super rich used to be able to have. Then the ordinary rich get them, and then the affluent get them, and it percolates on down."

"I think of wedding planners. When I was a kid, you had to be a Rockefeller or a movie star for a wedding planner. Now, that profession is a fixture in people's lives."

"In the future, the average person will have the same access to a great lawyer, or at least the abilities of a great lawyer at an affordable price."

"I wouldn't be surprised in the future if everyone had their own personal stylist."

Hudson Institute's @wrmead argues America's expert class is losing its authority because the world is changing faster than its institutions can adapt.

"We're now in an age where nobody knows the answers."

"The university culture, the intellectual culture has become much more wonkish, much less the creative innovator, much less the deep-souled investigator."

"With AI, who really knows what we're going to be dealing with in five years? Nobody."

"Who knows how to solve the problems of a technology that nobody understands? Nobody."

"Who knows how to manage the geopolitical competition between the United States and China? There's not a rulebook for much of this stuff."

"When waves of change come through, the experts aren't as good as they used to be. They don't know the answers."

"And as individuals, many of them just don't have the experiences or the personal qualities to deal wisely and effectively with the uncertainties that they face."

Unavoidable @tbpn @DylanAbruscato 👏🏿

Happy Wednesday. On today's show:

- @wrmead (Hudson Institute)
- @KatColeATL (AG1)
- @AarushSelvan (OpenAI)
- @realtarek (Stuut)
- @scottgmorton (Revel)
- @MeyerJosiah (Healthleap)
- Tyler Niday (Bonsai Robotics)

See you on the stream.

Some Math Problems, Scott Morton Breaks His Silence, Arsenal-2, Porsche Mission S, Walter Russell Mead Live x.com/i/broadcasts/1…