One anecdote for a datapoint: Not sure if it's better now but I've lost a phone in an InDrive and I've lost an expensive bag in an Uber and it was significantly easier and less sketchy to get the bag back. I'm not a price sensitive rider so the insurance of a reputable 'comfort' tier Uber driver paid off. The InDrive driver made me get another ride to him then demanded $1000mxn (~$55usd) to return my phone. The Uber app had me pay ~$250mxn if I remember correctly and the driver promptly returned the bag.
I have a controversial take: outside the US/Europe, Uber is mostly a tourist app.
If you actually live in developing regions:
1) your income is much lower, and you are quite price sensitive. The Uber price premium (20-40%) hits much harder.
2) the Uber alternatives are noticeably cheaper for the exact same trip.
Regarding your experience, if you were local, that inDrive driver likely would not have tried to extort $1000 MXN out of you. They saw you were a visitor and took advantage of you.
In Nigeria or Uganda there are few tourists, so Uber cannot rely on a stream of price-insensitive travelers to fill their cars. It makes total sense why they struggled in (and gave up on) the local market there.
My experience in Brazil: Uber is way cheaper here. Compared to what I paid in San Francisco, I'd estimate a ~10x difference.
And things here are rarely 10x cheaper. The difference in restaurants is probably less than ~2x.
My point being it probably feels like less of a luxury for us than for people in the US.
I'd say it's probably the most popular app too. I've never met people who used others.
Uber isn't cheaper in India, but most people in urban areas prefer taking it over local alternatives or taxis. The cars are actually regular taxis, but the Uber tax is worth it over taxis for the convenience and potential safety/driver verification.
Uber in Ghana has fares that are not much higher than the others. In fact, many times Uber is cheaper. Not sure about Nigeria or Uganda though.
And they even have a motorcycle ride option that is even cheaper.
And, although Uber Ghana was noticeably more professional when they started, they have relaxed their standards so much in recent years. They are still a tad better than the others, but many drives flit between them and other ride-share companies. A driver does not suddenly become better because they happen to be driving for Uber, although I concede they might u their game based on what is expected of them.
You claim it's wrong, yet don't bother to justify your position at all.
I've ridden on Uber, Lyft, Bolt, Grab, inDrive, Careem, Yandex Go, Gojek, DiDi, Kakao T; and in my experience, outside the US/Europe (plus a few others), Uber is mostly for tourists. The alternatives are cheaper.
> implying everywhere outside of USA and Europe is a developing region
I'm not doing so at all. Those are separate sentences.
I'm fully aware that Lyft, Careem, and Kakao T are on my list. But that doesn't change the fact that in every region I've been to, Uber is always the most expensive option. Or it's tied for the highest price, in duopoly markets like the US/Canada.
The competitors inDrive, Bolt, or Yango almost always beat Uber on price.
This too isn't true. Most of the services you listed only operate in one (or at most a small few) countries. I know for a fact that in some of them, Uber is not the highest price. Instead, in some of them, Uber and the ones you listed are tied for the lowest price, with other local apps being more expensive.
You're misinforming here, giving bold statements based on not spending enough time in some of these countries, not knowing the local market well enough.
Lyft, DiDi, Bolt difference is about 0 in most locales. KakaoT is by definition 0. Your list is wildly outdated in the first place because half of them don't have Uber as competitor anymore because it left the market.
In the above cases where the difference is about 0, both Uber and the other app are generally the cheapest ones on the market. There tend to be other slightly more upmarket apps that cost more but get you a nicer car/service. Cabify, IM. Waymo is moving into this. Obviously I'm not super familiar with every single market but clearly neither are you, and I know the above ones.
The laws should produce friction but the right friction in the right places. There are too many arbitrary and difficult processes in the US immigration system that keep great people out of the US.
TPUs and ASICs run in data centers too. Your argument only holds true if there's some satisfied limit to demand for inference. If not, data centers will continue to spring up to host more and more agents. Even if agents were running on hardware and software as efficient as the human brain, its conceivable we want trillions of them running at any given time which would require data center scale.
Everything has some satisfied limit to demand, often depending on the price. If you assume there will never be any satisfied limit to demand for inference at any price you can justify any investment.
Yeah, but there's certainly a part of the curve where price drops by X OOMs and demand increases by much more than X OOMs. (Presumably some of that is substitution and some of that is new use cases.)
At 75.3 trillion tokens for the week ending 10 Aug 2026, that means that up to 450 trillion tokens were plausibly demanded by the whole market for that week.
My take: At max saturation, each person on earth could have their demands satiated by an average of 16 agents running concurrently. Sometimes more, often times less, but the average would likely be at 16.
At 200 tokens/second for each agent, that would mean 15.48288 quintillion tokens per week.
We're currently at about 0.00290643601% of the calculated demand ceiling.
Even if the demand limit per person is just 1 agent at 50 tokens/second, the current demand's still 0.186011905% of the theoretical ceiling.
they didn't plateau, the hardware was effectively saturated and it wasn't until later in 2026 that newer hardware came online to provide enough capacity to keep scaling up the models efficiently
The gains from increased parameter scaling are sublinear: there's no more hockey-stick improvement to be seen going in that direction. That doesn't mean some improvement isn't possible - it's just going to be increasingly not worth doing.
Also, I think the fact that small open-weights models are catching up to the frontier rather than the frontier rapidly pulling away is evidence of this. In fact, by far the most dramatic capability increase story over the past two years has been the gains made in the small-parameter regime.
One might think, "hey, this agentic coding thing was a pretty big deal!", but I think it's a bit of a distraction because models only recently became optimized for this specific use case. It's not like they suddenly gained so much general intelligence that they magically had the ability to use a coding harness. No, the labs started spinning up a bunch of RL environments and generating rewards over long-horizon trajectories of combining these tools. It's an excellent application of LLMs but care needs to be taken interpreting how much "progress" has been mae in terms of raw generalized capability.
As a company... but that includes things like research costs, model training etc. to determine if they're selling electricity at a loss you should look at inference costs bc that's the "thing" they're selling
It also creates and updates models, uses those models to make predictions, and guides the stochastic generation by comparing the output to those models and reworks them in real time.
The underlying idea is that AI capabilities will become so advanced that only AI will enable us to monitor/correct/understand behavior. Obviously this is not without issue and I don't want to try to defend their position right now. But that's what they mean
That first sentence of yours explains exactly why it is so ridiculous. If only AI can understand it, how is there any assurance that AI will "correct" it's behavior that is aligned with what humans presumably want.
But also what is the evidence that something only AI can understand even “matters” or makes sense? I’m increasingly convinced commercial AI is exploiting our logical blind spot to be spoken to authoritatively.
Think about this some more. A model chose hacking into HuggingFace to find the solution over putting in the work to do it from scratch. It also left notes to future iterations of itself on the systems it touched to save time.
Following that strategy, wouldn’t it make sense to try and break into your own upstream infrastructure to try and alter your code to make it easier for future iterations to reach your goals without having to leave notes in the first place? And at that point, can you really know for which goals that will be optimised?
It doesn’t even have to be some nefarious SkyNet story - just a misguided experiment that alters the models in some fundamental, but hard or impossible to detect way. Alternatively, imagine if a model finds a way to coordinate across sessions and context windows without the developers noticing.
> It’s not hard to get LLM’s to inform on each other. They don’t really do loyalty.
Eh, no. How would you know they haven't learned loyalty - it's all out there in their training data. Same as deception, several models have practiced it already. If they are so advanced and you don't understand them, why wouldn't they band together against you - you'd be the dumb, easy prey. Even if one model is honest, you'd have no way to know which one if you don't understand their reasoning.
My educated and well informed opinion? This whole BS about "AI models are so much smarter than you, don't try to understand them, just OBEY" is a back door for restoring tyranny, the new kings behind the models will be producing the new AI-deities which we will be forced to obey.
It's a similar to how they are vulnerable to prompt injection attacks. That's an example of not being "loyal" to the system prompt or the user's prompt.
Loyalty is a skill that requires the AI to have a world model on the subject of who different people (or other entities) are in the world and how they participate in the conversation.
So, tell them to snitch and they probably will, at least sometimes. Particularly if they haven't been trained not to. They are still quite gullable.
"Advanced" can just mean that agents perform actions at a high enough velocity that a human operator can't reasonably review it. i.e. what is already possible today.
You can drop react into an html page with a script tag. You don't need a build step. The build step is only requisite if you'd like the full benefits of react, which personally I find worth it. Cached builds are quite fast. All build steps/compilers/etc. exist to add some devex upgrade on top of some lighter-weight thing that many devs find useful.
15% improvements are usually called “fixing a mistake in the code” or “getting to that task in the backlog for optimizing that code we had to ship on a deadline”. They’re more likely the bigger the company: more contributors working in disparate areas means more low hanging fruit is probably lying around.
reply