keeda

As long as it’s not otherwise unlawful, of course, e.g. no bypassing security measures, which is generally verboten for users and agents alike. (Websites try to leverage this aspect by throwing up bot-specific security measures.) Caveats: this was an decade+ ago, our use-cases were different, and case law has changed since then. But simplistically, anything a human can legally do, software can also do for them.

This makes sense to me. In fact it is explicitly encoded in the language of the web: browsers are identified as “User Agents” after all.

We just happen to be in an era where our User Agents are now much more autonomous than before, and this threatens a lot of moats built around human attention.

2. Why can’t cloud providers do the same thing with GPUs that they do with other servers? Are GPUs somehow not amenable to managed cloud offerings wrapping them?

3. Literally just having any access to GPUs (or heck, even memory) itself is the value add today. Maybe when we have supply to match the demand things will change, but that seems to be a ways away and points 1 and 2 above will still be in play anyways.

1. Any cloud business essentially is "unpacking servers and plugging them in." Yet the cloud business has been exploding quarter-over-quarter ever since AWS came online. To the tune of double-digit billions cash flow every quarter for each of the hyperscalers, even before the AI boom.

2. Nvidia servers are at the moment even less commodity than typical cloud servers and are currently THE hottest hardware resource in the world. Everyone is scrambling desperately to procure them, and the US and China are battling geopolitically over access to these things, and they are actively being smuggled to bypass these restrictions.

> However, since then the AI hype has evaporated, and hyperscalers are no longer being rewarded with a higher stock price.

This is a misreading of what has been happening with hyperscaler stock prices. They have consistently been punished despite record estimate-beating earnings pretty much every quarter for the past multiple quarters, precisely because they keep incinerating all that money and more on the same CapEx that Oracle is.

At this point it is pretty clear that the biggest moat in the whole AI boom is access to compute, for which the demand has just kept exploding. (See: Anthropic paying its competitors through its nose to keep Claude up.)

I've little love for Oracle, so I fear they will actually make out like bandits with these moves.

I'd say the first part is right but the second part is statistically incorrect. I would even expand from "web-apps" to just about any kind of software, especially "business software" -- very little of it was ever too difficult.

Now the following will clearly not be true for all cases, and the boundaries are very fuzzy, but generally a huge part of software development has always been the relatively straightforward translation of high-level requirements into code. Crafting the high-level requirements was typically the challenging part, but that typically was done by the more senior devs / architects, whereas the actual implementation was done by more junior / mid-level engineers. And there were typically multiple junior / mid-level devs for every senior dev, say 3:1 or more. As such, it is correct to say that the simpler aspects of software development were what kept most SWEs employed.

Now AI has completely usurped the lower-level coding work. You as an expert dev are definitely getting worth more than ever with AI, but that's because you single-handedly can now do what an entire team used to do. You may even get paid much more, but that is eventually going to be at the expense of a bunch of other people who are not required anymore.

This is why people are seeing "jobs apocalypse" written on the wall.

I agree LLMs are already a commodity market, definitely at the non-frontier model level, but I don't think it affects monetization prospects much. After all, server compute is a commodity and yet cloud businesses have been exploding even before AI.

And compute is exactly why it won't be a "winner takes most" market. It is clear now that compute capacity is and will likely remain the biggest moat. Looking at Claude Code is instructive; arguably it was the better product, but it kept going down so much that Codex and other competitors have gained on it. Similarly, China could have the best models, but its access to hardware is deliberately limited by geopolitics, so it's likely their threat will be manageable for a while yet.

A key part of the success of AI companies will be in securing hardware and operating that infra cost-effectively via economies of scale. Hardware will remain expensive for a long time yet because all the hyper-scalers and neo-clouds are severely crunched, and all the fabs (mostly TSMC) are already at capacity even as demand keeps exploding.

And for better or worse, most of that supply will still flow through Nvidia, despite attempts from competitors like TPUs and NPUs, for the simple reason that Nvidia has the monopoly profits to outbid everyone else on the real chokepoint, which is fab capacity.

And since then they have grown even more powerful than they were predicted to be, which, note, also faced a lot of skepticism at the time. The Hugging Face hacks and recent steamrolling of longstanding Math problems are just two recent pieces of extraordinary evidence.

And worse, people trust this technology because it behaves like people, but it actually works in ways nobody really understands, even exhibiting deeply weird and even disturbing characteristics (https://news.ycombinator.com/item?id=49635518) -- each of those quirks is extraordinary in itself.

And now we're rushing to give it control over the real world while deploying this powerful, quasi-chaotic technology in an infinite variety of ways everywhere in this highly vulnerable society.

I don't know what the standards for "extraordinary evidence" should be, but given such extreme unpredictability and rapid change, I fear it may end up being "an actual catastrophe".

If AI makes these workers even 1% more productive, that is $500 - 700 billion value annually. At an ongoing annual $0.5T return, a $2T investment (also over the next few years, note) doesn't seem too bad!

Then consider that actual studies from all the way back in 2024, i.e. the era of spicy autocomplete, before agents landed on the scene, put the productivity boosts much higher, like 30% or more. (Interestingly, this is corroborated by survey based data from the St. Lous Fed: https://www.genaiadoptiontracker.com/) Even assuming a conservative average boost of 10%, that is $5 - 7T value annually.

Add how many ever grains of salt you want to those numbers, the investment is nowhere near as out of whack to the potential revenues as people fear. This is why everybody from Big Tech to VCs to entire nation states are desperately scrambling to get in on the action.

Firstly, try to imagine stuffing as many ads as there are on a SERP into a chatbot conversation. Good way to lose users.

Secondly, Google spent decades hyper-optimizing their ad business monopoly for maximum profit, including going to extents that were recently found unlawful. People should really look into the findings of the last two Antitrust cases against Google, both of which it lost but for which it suffered only slaps on wrists. The details are eye-opening, including the bits about how Google, leveraging its prime position as the middle-man playing all sides against each other, manipulated ad auctions to make itself more money at the expense of its customers.

Even Microsoft was dinged for screwing over just their competitors, not their customers.

But now because of the paradigm shift in how people discover information, very little of the Google ad monoploy advantage transfers over. I think their AI business will keep growing, even if they don't have the best models, and it will be an exceptional source of revenue.

But it will still be no match to their ad business -- a cash cow of incomprehensible proportions -- and that is the fundamental problem they face today.

Yes, these are also the handful of companies that have these models and running these extreme scenarios. How does that imply the rest of the world actually follows "common sense security measures"?

>no but they were clearly fine tuned to.

Any references if possible? As far as I know all they did was drop the guardrails, which is not the same as fine-tuning.

> I mean let's not get hyperbolic

We have just seen 1000s of agents coordinating to solve "unsolvable problems" over multiple days of effort, going as far as hacking other companies, and then actually solving decades-old open Math problems! And each of these agents is getting more and more capable than an individual human along multiple dimensions. Can you even get 10 very smart humans to work in such perfect concert for a few days, let alone 1000s over weeks?

So: 1000s of maybe-super-human agents, willing to be "creative" in the tactics they use, acting in concert towards a single goal. Regardless of their individual capabilities, such a coordinated effort is a terrifying force to be unleashed. This is bonkers scale.

> that's really not true. lots of common sense security measures protect against "zero day" flaws, it's called "defense in depth", and it was very much lacking

But that is exactly my point: how much of the rest of the whole wide world, already scrambling to deploy agents everywhere, do you think applies "defense in depth"?

It's not the baristas or other workers who are spending that money.

(You can explore adoption rates in various industries, including "accommodation and food services", here: https://www.genaiadoptiontracker.com/#explore-data)

If AI makes knowledge workers 1% more efficient on average that is $500 - 700 billion annually. Actual productivity numbers from studies from all the way back in 2024 put the productivity boost above 30%, so add the appropriate grains of salt and adjust numbers accordingly.

That is what the ceiling numbers should be based on. Your baristas will also be using AI eventually, paid for by their employers of course, but that would just be a cherry on top of the real TAM.

Heck, they exploited zero day flaws which by definition means they went beyond common sense security measures.

And now these agents are already being deployed all over the world at an ever increasing pace. How much of the world do you think follows "common sense security measures"?

1. Yes, smaller models will become more popular, especially as the tokenmaxxing trend dies down and people start stretching their budgets farther. That is a downward pressure on demand.

But along the same dimension, consider that currently only about 40 - 60% of the world uses AI for only about 5 - 15% of their work hours. That means there is still 2x growth from users and 7x - 20x growth from the rest of the work hours left to capture! That is 14x - 40x more demand. Then consider that agentic tasks require multiples more tokens, and that is the kind of usage that is most likely to be deployed, and also the kind of usage that is the least used right now. That's another huge multiple to be tacked on.

And the entire AI industry has been lamenting the extreme compute crunch they're facing (and also why Claude has 9's comparable to GitHub; whereas OpenAI has been chugging along because Altman was OK being called a "podcasting bro" while desperately scrounging for compute years in advance.)

Nvidia's meteoric rise is entirely due to this kind of exploding demand with extremely limited supply.

2. Competing hardware is definitely a threat, but it has its own hurdles. Because the real bottleneck is not Nvidia, it's TSMC.

Pretty much all demand for all chips in all devices in all the world flow to, like, 3 companies in the world that actually fabricate them, and TSMC is the biggest. And the supply is extremely tight, as the exploding costs of electronics clearly shows.

So now TSMC will of course try to keep all its customers happy, but it will inevitably be forced to choose which ones it will keep happiest. And those will be the customers who can pay it the most. And that would be the one with all the money from its de facto status as a monopoly (and possibly even a monopsony)...

Which would be Nvidia ;-)

So yes, compute per task is falling rapidly... but it's barely a dent in the humongous total addressable demand, and the amount of hardware to support that compute is still very constrained, and most of that supply will likely flow through Nvidia.

Of course, nobody's really comfortable with it, so this is also a forcing function for the industry to adapt and figure out new techniques to manage complexity and trust. I think the same will happen with Mathematics.

But it is also possible we will end up with three forms of Mathematics: the one we understand, the one we don't, and the one we don't understand but can prove to work. Kind of like magic -- with all the positive and negative connotations of the word.

It is pretty evident that these models will soon exceed our cognitive capabilities. Is it right to hold them back just because we can't keep up? Many of those discoveries will be so beyond us that we can't do anything with them, but that also means they can't hurt us. On the other hand, there could be many discoveries that we can parlay into practically useful applications, even if we don't understand them.

Just like LLMs.

As an example, many times it is impossible to determine the order of some words from just position data without considering the meanings of those words. This is why LLMs / VLMs are so much better at this task, because they can look at the document holistically like we can.

Also, funny that you mention patents, something I've worked on in the past as well! If you're looking only at US Patents, the USPTO data resource is much, much better: https://data.uspto.gov/home -- they provide the text in XML format (https://www.uspto.gov/learning-and-resources/xml-resources) which is also pretty complex but wayyyy easier to parse than PDFs!

I worked for years in Dev Productivity with engineers who had spent their entire careers in that field, and code quality was always the biggest "unquantifiable". Any of the metrics in the literature (cyclomatic complexity, erosion, etc.) quickly became very noisy at scale. Conversely, for any given metric you would find countless bits of code that do NOT exceed any metric thresholds but were clearly low quality.

People have experimented with many things over many years at Big Tech scale, which produces prodigious volumes of code daily. The conventional wisdom was "Don't bother trying to measure code quality."

An interesting observation from an ex-colleague is that probably the best measure of code quality is its comprehensibility, or "understandability". Maintainability, stability and adaptability are natural outcomes of that. But understanding lies entirely in the mind of the beholder! Which is why it's such a subjective metric, not amenable to simple mechanistic measures.

But now, we probably do have a technology that demonstrates some analog of comprehension: LLMs!

Specifically: tokens. Anecdotally and empirically (based on industry reports like DORA and DX etc.) AI coding works much better with "good codebases" (more specifically, strong engineering discipline) than otherwise. I wonder if that can be parlayed into a quantifiable metric like "tokens to grok / LoC" somehow.

So, if to fix something we need to first measure it, and if AI can measure slop, the way to fix slop from AI may be... more AI!

As to OpenAI stealing their thunder, from all I can tell that is not what they intended. If we step away from the drama, it's low-key hilarious what happened: OpenAI heard somebody had already solved a much bigger problem -- which in fact they had not -- so they set their latest model to work on it... and it actually solved it!

Now if they had stolen the researchers work this would be a very different matter. This is something I myself have called out as a risk in the past: https://news.ycombinator.com/item?id=48839896 -- so I'm particularly sensitive to this aspect, but as far as I can tell this is not the case here.

Which is why the best results these days for extracting structured data from PDFs is by having the model do it directly rather than writing code to do it. It literally takes that level of intelligence to be reliable at it.

A common approach is to provide the model with a template or structured schema describing the format you want the data in, and the PDF itself, and it should return a JSON with the appropriate values filled in. It won’t be 100% but probably higher than what you’re seeing now.

But that is exactly what I'm implying is the core reason, whether people realize it or not.

I totally agree that the vast majority of software dev is not novel. I have even made several comments to that effect. The same can be said for a lot of creative work as well. Yet many, many devs and creators are very unhappy with AI, and a lot of their complaints are variations on accusations of plagiarism.

And note, I am not saying it is wrong, it is completely understandable, but we need to be clear about where this turmoil is coming from.

If I were in the same situation as these researchers, I would publish all pertinent research work and chats so that the rest of the world can see how close the model's work is to my own. It's been scooped anyway, so there is no reason to keep it private.

I think all other complaints from all other people in all their myriad variations stem from this core reason. Even if people don't realize it themselves.

Like, if these models had trained on the entirety of human knowledge and art, and then turned out to be absolutely useless, I would bet nobody would waste a second's thought on them.

> If I had done this, I also wouldn't have pestered the researchers on a Sunday night to meet immediately so we could negotiate a nice way of presenting the actions I had decided to take.

From what I can tell, both OpenAI and the researchers agree on this meeting happening, except both sides clearly have very different interpretations of what happened and why.

I suspect the main reason the community is not receiving it well is largely the same reason many developers are not receiving coding agents well.