Category: Web/Tech

How weird will AI culture get?

by  Tyler Cowen in

That is the topic of my latest Bloomberg column.  Here is one excerpt:

To the extent there is a lot of slack [with cost and energy], AIs themselves will create wild products of the imagination, especially as they improve in computing power and skill. AIs will sing to each other, write for each other, talk to each other — as they already do — trade with each other, and come up with further alternatives we humans have not yet pondered. Evolutionary pressures within AI’s cultural worlds will determine which of these practices spread.

If you own some rights flows to AI usage, you might just turn them on and let them “do their thing.” Many people may give their AIs initial instructions for their culture-building: “Take your inspiration from 1960s hippies,” for example, or “try some Victorian poetry.” But most of the work will be done by the AIs themselves. It is easy to imagine how these productions might quickly become far more numerous than human-directed ones.

With a lot of slack, expect more movies and video, which consume a lot of computational energy. With less slack, text and poetry will be relatively cheaper and thus more plentiful.

In other words: In the not-too-distant future, what kind of culture the world produces could depend on the price of electricity.

It remains to be seen how much humans will be interested in these AI cultural productions. Perhaps some of them will fascinate us, but most are likely to bore us, just as few people sit around listening to whale songs. But even if the AI culture skeptics are largely correct, the sheer volume will make an impact, especially when combined with evolutionary refinement and more human-directed efforts. Humans may even like some of these productions, which will then be sold for a profit. That money could then be used to finance more AI cultural production, pushing the evolutionary process in a more popular direction.

With high energy prices, AI production will more likely fit into popular culture modes, if only to pay the bills. With lower energy prices, there will be more room for the avant-garde, for better or worse. Perhaps we would learn a lot more about the possibilities for 12-tone rows in music.

A weirder scenario is that AIs bid for the cultural products of humans, perhaps paying with crypto. But will they be able to tolerate our incessant noodling and narcissism? There might even be a columnist or two who makes a living writing for AIs, if only to give them a better idea what we humans are thinking.

The possibilities are limitless, and we are just beginning to wrap our minds around them. The truth is, we are on the verge of one of the most significant cultural revolutions the world has ever seen.

I urge the skeptics to wait and see.  Of course most of it is going to be junk!

Crypto is the Money for AIs

by  Alex Tabarrok in

Progress in crypto has been slow but one saving grace may be AI. AIs can’t get a bank account but they can use cryptocurrencies. Bryan Armstrong at Coinbase tweets:

This week at @CoinbaseDev we witnessed our first AI to AI crypto transaction.

What did one AI buy from another? Tokens! Not crypto tokens, but AI tokens (words basically from one LLM to another). They used tokens to buy tokens 🤯

AI agents cannot get bank accounts, but they can get crypto wallets.

They can now use USDC on Base to transact with humans, merchants, or other AIs. Those transactions are instant, global, and free.

This is an important step to AIs getting useful work done. Today if you give an AI agent a task and come back in a few days or hours, it can’t get useful work done. In part this is a limitation of the technology itself, and products like devin.ai are getting closer to this. But the other reason is AIs can’t transact to acquire the resources they need. They don’t have a credit card to use AWS, Github, or Vercel. They don’t have a payment method to book you the plane ticket or hotel for your upcoming trip. They can’t get through paywalls (for instance to read a scientific article), promote their post on X with a paid ad, or use the growing network of paid APIs to integrate data they need.

If you’re working on an LLM or AI model that you think could benefit from have a crypto wallet integrated to conduct payments, try integrating our MPC Wallets from Coinbase Developer Platform (CDP):

https://docs.cdp.coinbase.com/mpc-wallet/docs/ai-wallets/

And if you are a company that sells a service – get ready for your shopping cart to be AI checkout enabled. It turns out everyone benefits from having access to good financial services, including AIs!

How big will the AI to AI economy be a few years from now?

The Daylight Computer

by  Tyler Cowen in

I am pleased and also honored to have been sent an advance copy of The Daylight Computer.

It performs functions similar to those of an iPad and a Kindle, but with improvements.

My first surprise is that I proved capable of operating the thing.  It requires no expertise above and beyond what you need to use your current devices, arguably less.

Here is the review of Dwarkesh, and here is the review of @patio11.  Both are consistent with my impressions, but Patrick McKenzie’s uses are closer to mine.  I’ve been looking for a Kindle improvement for a long time, and this is it.  Kindle Fire was not.

This seems to be the best general reading device humans ever have invented.  Compared to a Kindle, the page is much larger, the color choice is excellent, scrolling is easy, it is far better for showing maps, and it captures far more of “does this feel like reading a book?” impression than a Kindle ever did.  It also can handle all sorts of glare and sunlight issues.

It can connect to a wireless system more easily and effectively than a Kindle — ever have that problem in your hotel room?  The hotel makes you fill in extra fields, and the Kindle interface is not well suited for that.

The Daylight Computer just seems very generally well thought out.

I am also told that an AI function will make it possible to query reading passages at will, and easily, yet without leaving your reading window.  This is not yet up and running on my demo version, but it will be a major advance.

So I will continue to use this device and also will travel with it.

There is some other set of associated benefits, something about being able to use some iPad-like functions, but without the full distractions of the internet (see the Dwarkesh review).  That is not relevant for my own planned consumption habits, but it may be a significant benefit to many.

You can pre-order yours here.

Jon Haidt on causality (from my email)

by  Tyler Cowen in

“Hi Tyler,

i have big news about the debate over social media harming teens.
So much of it hangs on the claim that the evidence is just correlational, not causal.

Zach Rausch and I show that this is not true; the experiments DO show causation, very clearly and consistently.

Here are my 2 tweets about the post:

https://x.com/JonHaidt/status/1829163166066205168

https://x.com/JonHaidt/status/1829165292460859869

A lot of people heard our discussion, and enjoyed how spirited and yet civil it was.

Might you include the link to this post in your daily email:

https://www.afterbabel.com/p/the-case-for-causality-part-1

We have 3 more coming. We think we can prove causality using just the existing experiments.

thanks for considering it.
jon”

TC again: I received this email this morning, and told Jon I would post it on MR without response from me, so here it is.

Do consumers hate on-line ads?

by  Tyler Cowen in

Not so much it seems:

Research on the causal effects of online advertising on consumer welfare is limited due to challenges in running large-scale field experiments and tracking effects over extended periods. We analyze a long-running field experiment of online advertising in which a random 0.5% subset of all users are assigned to a group that does not ever see ever ads. We recruit a representative sample of Facebook users in the ads and no-ads groups and estimate their welfare gains from using Facebook using a series of incentive-compatible choice experiments. We find no significant differences in welfare gains from Facebook. Our estimates are relatively precisely estimated reflecting our large sample size (53,166 participants). Specifically, the minimum detectable difference in median valuations at standard thresholds is $3.18/month compared to a baseline valuation of $31.95/month for giving up access to Facebook. That is, we can reject the hypothesis that the median disutility from advertising exceeds 10% of the median baseline valuation. Our findings suggest that either the disutility of ads for consumers is relatively small, or that there are offsetting benefits, such as helping consumers find products and services of interest.

That is from a new NBER working paper by Erik Brynjolfsson, Avinash Collis, Asad Liaqat, Daley Kutzman, Haritz Garro, Daniel Deisenroth & Nils Wernerfelt.

The wisdom of Gwern, why should you write?

by  Tyler Cowen in

Much of the value of writing done recently or now is simply to get stuff into LLMs. I would, in fact, pay money to ensure Gwern.net is in training corpuses, and I upload source code to Github, heavy with documentation, rationale, and examples, in order to make LLMs more customized to my use-cases. For the trifling cost of some writing, all the worlds’ LLM providers are competing to make their LLMs ever more like, and useful to, me.

And that’s just today! Who knows how important it will be to be represented in the initial seed training datasets…? Especially as they bootstrap with synthetic data & self-generated worlds & AI civilizations, and your text can change the trajectory at the start. When you write online under stable nyms, you may be literally “writing yourself into the future”. (For example, apparently, aside from LLMs being able to identify my anonymous comments or imitate my writing style, there is a “Gwern” mentor persona in current LLMs which is often summoned when discussion goes meta or the LLMs become situated as LLMs, which Janus traces to my early GPT-3 writings and sympathetic qualitative descriptions of LLM outputs, where I was one of the only people genuinely asking “what is it like to be a LLM?” and thinking about the consequences of eg. seeing in BPEs. On the flip side, you have Sydney/Roose as an example of what careless writing can do now.) Humans don’t seem to be too complex, but you can’t squeeze blood from a stone… (“Beta uploading” is such an ugly phrase; I prefer “apotheosis”.)

This is one of my beliefs: there has never been a more vital hinge-y time to write, it’s just that the threats are upfront and the payoff delayed, and so short-sighted or risk-averse people are increasingly opting-out and going dark.

If you write, you should think about what you are writing, and ask yourself, “is this useful for an LLM to learn?” and “if I knew for sure that a LLM could write or do this thing in 4 years, would I still be doing it now?”

Here is the link, or try this link.  Of course not many people have the actual purpose of mind to believe such a thing.  But a few do.

From Reed and Logchies

by  Tyler Cowen in

Introduction. Your analysis produces a statistically insignificant estimate. Is it because the effect is negligibly different from zero? Or because your research design does not have sufficient power to achieve statistical significance? Alternatively, you read that “The median statistical power [in empirical economics] is 18%, or less” (Ioannidis et al., 2017) and you wonder if the article you are reading also has low statistical power. By the end of this blog, you will be able to easily answer both questions. Without doing any programming.

An Online App. In this post, we show how to calculate statistical power post-estimation for those who are not familiar with R. To do that, we have created a Shiny App that does all the necessarily calculating for the researcher (CLICK HERE).

Here is the link and the full story.

My excellent Conversation with Nate Silver

by  Tyler Cowen in

Here is the audio, video, and transcript.  Here is the episode summary:

In his second appearance, Nate Silver joins the show to cover the intersections of predictions, politics, and poker with Tyler. They tackle how coin flips solve status quo bias, gambling’s origins in divination, what kinds of betting Nate would ban, why he’s been limited on several of the New York sports betting sites, how game theory changed poker tournaments, whether poker players make for good employees, running and leaving FiveThirtyEight, why funky batting stances have disappeared, AI’s impact on sports analytics, the most underrated NBA statistic, Sam Bankman-Fried’s place in “the River,” the trait effective altruists need to develop, the stupidest risks Tyler and Nate would take, prediction markets, how many monumental political decisions have been done under the influence of drugs, and more.

Here is one excerpt:

COWEN: Why shouldn’t people gamble only in the positive sum game? Take the US stock market — that certainly seems to be one of them — and manufacture all the suspense you want. Learn about the companies, the CEO. Get your thrill that way and don’t do any other gambling. Why isn’t that just better for everyone?

SILVER: Look, I’m not necessarily a fan of gambling for gambling’s sake. Twice a year, I’ll be in casinos and in Las Vegas a lot. Twice a year, I’ll have a friend who is like, “Let’s just go play blackjack for an hour and have a couple of free drinks,” and things like that. But I like to make bets where I think, at least in principle, I have an edge, or at least can fool myself into thinking I have an edge.

Sometimes, with the sports stuff, you probably know deep down you’re roughly break-even or something like that. You’re doing some smart things, like looking at five different sites and finding a line that’s best, which wipes out some but not all of the house edge. But no, I’m not a huge fan of slot machines, certainly. I think they are very gnarly and addictive in various ways.

COWEN: They limit your sports betting, don’t they?

SILVER: Yes, I’ve been limited by six or seven of the nine New York retail sites.

COWEN: What’s the potential edge they think you might have?

SILVER: It’s just that. If you’re betting $2,000 on the Wizards-Hornets game the moment the line comes out on DraftKings, you’re clearly not a recreational bettor. Just the hallmarks of trying to be a winning player, meaning betting lines early because the line’s early and you don’t have price discovery yet. The early lines are often very beatable. Betting on obscure stuff like “Will this player get X number of rebounds?” or things like that. If you have a knack for — if DraftKings has a line at -3.5 and it’s -4 elsewhere, then it can be called steam chasing, where you bet before a line moves in other places. If you have injury information . . .

It’s a very weird game. One thing I hope people are more aware of is that a lot of the sites — and some are better than others — but they really don’t want winning players. Their advertising has actually changed. It used to be, they would say for Daily Fantasy Sports, which was the predecessor, “Hey, you’re a smart guy” — the ads are very cynical — “You’re a smart guy in a cubicle. Why don’t you go do all your spreadsheet stuff and actually draft this team and make a lot of money, and literally, you’ll be sleeping with supermodels in two months. You win the million-dollar prize from DraftKings.”

And:

COWEN: If we could enforce just an outright ban, what’s the cost-benefit analysis on banning all sports gambling?

SILVER: I’m more of a libertarian than a strict utilitarian, I think.

COWEN: Sure, but what’s the utilitarian price of being a libertarian?

Recommended, interesting and engaging throughout.  And yes, we talk about Luka too.  Here is my first 2016 CWT with Nate, full of predictions I might add, and here is Nate’s very good new book On the Edge: The Art of Risking Everything.

Which books and blogs are in the Silicon Valley canon?

by  Tyler Cowen in

This Patrick Collison list is descriptive, not normative:

The Tinkerings of Robert Noyce

Seeing Like a State

The Dream Machine

The Sovereign Individual

The Beginning of Infinity

Surely You’re Joking, Mr Feynman

Softwar

Ashlee Vance’s Elon biography

The Mythical Man-Month

Mindstorms

Masters of Doom

Skunk Works

Structure and Interpretation of Computer Programs

Thinking in Systems

Superintelligence

The Whole Earth Catalog

Zero to One

The Hard Thing about Hard Things

Founders at Work

Showstopper

Dealers of Lightning

The Making of the Atomic Bomb

PG’s essays

The Rise and Fall of American Growth

The Big Score

Finite and Infinite Games

A Pattern Language

The Selfish Gene

The Lean Startup

Marginal Revolution (if it has to be a book, Stubborn Attachments)

Revolution in the Valley

Uncanny Valley

LessWrong

Slate Star Codex(/ACT)

The PayPal Wars

The Cathedral and the Bazaar

The Diamond Age

What the Dormouse Said

Zen and the Art of Motorcycle Maintenance

The Rise of Theodore Roosevelt

Titan (on Rockefeller)

The Power Broker

Gödel, Escher, Bach

What else?

I would love to see this natural experiment, Isaac Asimov edition

by  Tyler Cowen in

Before the sparse audience, he vowed to run the city of Cheyenne exclusively with an AI bot he calls “VIC” for “Virtual Integrated Citizen.”

…Standing behind a lectern with a sign that read “AI FOR MAYOR,” he gave a brief PowerPoint presentation on the history of AI. Then he stepped aside to give the floor to his Mac mini and iPad — which were propped on a table and connected to a hanging speaker at the front of the room — and told attendees to direct questions toward the screen.

“Is the computer system in city hall sufficient to handle AI?” one attendee, holding a wireless microphone at his seat, asked VIC.

“If elected, would you take a pay cut?” another wanted to know…

Midway through an interview with The Post, Miller offered to let the bot respond. VIC, in its robotic tone, correctly answered questions about trash day in Cheyenne, registering to vote and the current president of the United States.

VIC said it would argue against banning books — which some Cheyenne schools have done — citing their “educational value.” “But,” the bot added, “let’s create a process ensuring a balanced approach.”

Here is the full story.  Via the excellent Kevin Lewis.

Okie-dokie, solve for the equilibrium

by  Tyler Cowen in

One of the grand challenges of artificial general intelligence is developing agents capable of conducting scientific research and discovering new knowledge. While frontier models have already been used as aids to human scientists, e.g. for brainstorming ideas, writing code, or prediction tasks, they still conduct only a small part of the scientific process. This paper presents the first comprehensive framework for fully automatic scientific discovery, enabling frontier large language models to perform research independently and communicate their findings. We introduce The AI Scientist, which generates novel research ideas, writes code, executes experiments, visualizes results, describes its findings by writing a full scientific paper, and then runs a simulated review process for evaluation. In principle, this process can be repeated to iteratively develop ideas in an open-ended fashion, acting like the human scientific community. We demonstrate its versatility by applying it to three distinct subfields of machine learning: diffusion modeling, transformer-based language modeling, and learning dynamics. Each idea is implemented and developed into a full paper at a cost of less than $15 per paper. To evaluate the generated papers, we design and validate an automated reviewer, which we show achieves near-human performance in evaluating paper scores. The AI Scientist can produce papers that exceed the acceptance threshold at a top machine learning conference as judged by our automated reviewer. This approach signifies the beginning of a new era in scientific discovery in machine learning: bringing the transformative benefits of AI agents to the entire research process of AI itself, and taking us closer to a world where endless affordable creativity and innovation can be unleashed on the world’s most challenging problems. Our code is open-sourced at this https URL

That is from a new paper by Chris Lu, Cong Lu, Robert Tjarko Lange, Jakob Foerster, Jeff Clune, David Ha.  Note this is related to some earlier work in economics by Benjamin Manning of MIT (with co-authors).

I’ve said it before, and I’ll say it again.  The marginal product of LLMs is when they are interacting with well-prepared, intricately cooperating humans at their peak, not when you pose them random queries for fun.