Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

AI is different

antirez.com

Post reply on HN
Print view View original

511–520 of 920 posts

Re: AI is different

#511

Post by yoz-y »

jhbadger wrote: post #425
keiferski wrote: post #411

There’s a simple flaw in this reasoning: Just because X can be replaced by Y today doesn’t imply that it can do so in a Future where we are aware of Y, and factor it into the background assumptions about the task. In more concrete terms: if “not being powered by AI” becomes a competitive advantage, then AI won’t be meaningfully replacing anything in that market. You can already see this with YouTube: AI-generated vid…

>You can already see this with YouTube: AI-generated videos are a mild amusement, not a replacement for video creators, because made by AI is becoming a negative label in a world where the presence of AI video is widely known. But that's because, at present, AI generated video isn't very good. Consider the history of CGI. In the 1990s and early 2000s, it was common to complain about how the move away from practical s…

That said, the complaint is coming back. Namely because most new movies use an incredible amount of CGI and due to the time constraints the quality suffers.

As such, CGI is once again becoming a negative label.

I don’t know if there is an AI equivalent of this. Maybe the fact that as models seem to move away from a big generalist model at launch, towards a multitude of smaller expert models (but retaining the branding, aka GPT-4), the quality goes down.

Re: AI is different

#512

Post by tomrod »

__MatrixMan__ wrote: post #431
jhbadger wrote: post #425

Earlier quoted context omitted.

>You can already see this with YouTube: AI-generated videos are a mild amusement, not a replacement for video creators, because made by AI is becoming a negative label in a world where the presence of AI video is widely known. But that's because, at present, AI generated video isn't very good. Consider the history of CGI. In the 1990s and early 2000s, it was common to complain about how the move away from practical s…

Do you get the feeling that AI generated content is lacking something that can be incrementally improved on? Seems to me that it's already quite good in any dimension that it knows how to improve on (e.g. photorealism) and completely devoid of the other things we'd want from it (e.g. meaning).

It's missing random flaws. Often the noise has patternd as a result of the diffusion or generation process.

Re: AI is different

#513

Post by andai »

keiferski wrote: post #411

There’s a simple flaw in this reasoning: Just because X can be replaced by Y today doesn’t imply that it can do so in a Future where we are aware of Y, and factor it into the background assumptions about the task. In more concrete terms: if “not being powered by AI” becomes a competitive advantage, then AI won’t be meaningfully replacing anything in that market. You can already see this with YouTube: AI-generated vid…

This only works in a world where AI sucks and/or can be easily detected. I've already found videos where on my 2nd or 3rd time watching I went, "wait, that's not real!" We're starting to get there, which is frankly beyond my ability to reason about.

It's the same issue with propaganda. If people say a movie is propaganda, that means the movie failed. If a propaganda movie is good propaganda, people don't talk about that. They don't even realize. They just talk about what a great movie it is.

Re: AI is different

#514

Post by mycall »

amelius wrote: post #460
thephotonsphere wrote: post #259

Earlier quoted context omitted.

Tesla uses only cameras, which sounds crazy (reflections, direct sunlight disturbances, fog , smoke, etc. LiDAR, radar assistance feels crucial https://fortune.com/2025/08/15/waymo-srikanth-thirumalai-int...

The nice thing about LiDAR is that you can use it to train a model to simulate a LiDAR based on camera inputs only. And of course to verify how good that model is.

I can't wait until V2X and sensor fusion comes to autonomous vehicles, greatly improving the detailed 3D mapping of LiDAR, the object classification capabilities of cameras, and the all-weather reliability of radar and radio pings.

Re: AI is different

#515

Post by andai »

siliconc0w wrote: post #500

I'm on team plateau, I'm really not noticing increasing competency in my daily usage of the major models. And sometimes it seems like there are regressions where performance drops from what it could do before. There is incredible pressure to release new models which means there is incredible pressure to game benchmarks. Tbh a plateau is probably the best scenario - I don't think society will tolerate even more inequa…

I think the current economy is already dreadful. So I don't have much desire to maintain that. But it's easier to break something further than to fix it, and god knows what AI is going to do to a system with so many feedback loops.

Re: AI is different

#516

Post by layer8 »

brap wrote: post #347
xg15 wrote: post #328

> Nor is it possible to imagine a system where a few mega companies are the only providers of intelligence Why not? This seems to be exactly where we're headed right now, and the current administration seems to be perfectly fine with that trend. If you follow the current logic of AI proponents, you get essentially: (1) Almost all white-collar jobs will be done better or at least faster by AI. (2) The "repugnant concl…

I find it extremely hard to believe that ASI will still require enormous investments in a post-ASI world. The initial investment? Likely. But there have to be more efficient ways to build intelligence, and ASI will figure it out. It did not take trillions of dollars to produce you and I.

The efficient ways (biotech?) are still likely to require massive investments, maybe not unlike chip fabs that cost billions. And then IP and patents come in.

Re: AI is different

#517

Post by jaynetics »

keiferski wrote: post #420
keiferski wrote: post #411

There’s a simple flaw in this reasoning: Just because X can be replaced by Y today doesn’t imply that it can do so in a Future where we are aware of Y, and factor it into the background assumptions about the task. In more concrete terms: if “not being powered by AI” becomes a competitive advantage, then AI won’t be meaningfully replacing anything in that market. You can already see this with YouTube: AI-generated vid…

Just to further elaborate on this with another example: the writing industry. (Technical, professional, marketing, etc. writing - not books.) The default logic is that AI will just replace all writing tasks, and writers will go extinct. What actually seems to be happening, however, is this: - obviously written-by-AI copywriting is perceived very negatively by the market - companies want writers that understand how to…

As someone who used to be in the writing industry (a whole range of jobs), this take strikes me as a bit starry-eyed. Throw-away snippets, good-enough marketing, generic correspondence, hastily compiled news items, flairful filler text in books etc., all this used to be a huge chunk of the work, in so many places. The average customer had only a limited ability to judge the quality of texts, to put it mildly. Translators and proofreaders already had to prioritize mass over flawless output, back when Google Translate was hilariously bad and spell checkers very limited. Nowadays, even the translation of legal texts in the EU parliament is done by a fraction of the former workforce. Very few of the writers and none of the proofreaders I knew are still in the industry.

Addressing the wider point, yes, there is still a market for great artists and creators, but it's nowhere near large enough to accommodate the many, many people who used to make a modest living, doing these small, okay-ish things, occasionally injecting a bit of love into them, as much as they could under time constraints.

Re: AI is different

#518

Post by paulryanrogers »

ACCount37 wrote: post #467
latexr wrote: post #377

Earlier quoted context omitted.

> Humans use only cameras. Not true. Humans also interpret the environment in 3D space. See a Tesla fail against a Wile E. Coyote-inspired mural which humans perceive: https://youtu.be/IQJL3htsDyQ?t=14m34s

This video proves nothing other than "a YouTuber found a funny viral video idea". Teslas "interpret the environment in 3D space" too - by feeding all the sensor data into a massive ML sensor fusion pipeline, and then fusing that data across time too. This is where the visualizers, both the default user screen one and the "Terminator" debugging visualizer, get their data from. They show plain and clear that the car op…

I've seen semi trucks with scenic views painted on them, both rear and side panels.

Re: AI is different

#519

Post by yladiz »

BoorishBears wrote: post #491
yladiz wrote: post #471

Earlier quoted context omitted.

I’m a bit confused. If we’re talking about consumer cars, the end goal is not to rent a car that can drive itself, the end goal is to own a car that can drive itself, and so it doesn’t matter if the car is available for purchase but costs $250,000 because few consumers can afford that, even wealthy ones.

a) I'm not talking about consumer cars, you are. I said very plainly this level of capability won't reach consumers soon and I stand by that. Some Chinese companies are trying to make it happen in the US but there's too many barriers. b) If there was a $250,000 car that could drive itself around given major cities, even with the geofence, it would sell out as many units as could be produced. That's actually why I tel…

If you look at the original comment that you replied to, the goalpost was explained clearly:

> The goalpost will be when you can buy one and drive it anywhere.

So let’s just ignore the non-consumer parts entirely to avoid shifting the goalpost. I still stand by the fact that the average (or median) consumer will not be able to afford such an expensive car, and I don’t think it’s controversial to state this given the readily available income data in the US and various other countries. The point isn’t that it exists, Rolls Royce and Maseratis exist, but they are niche and so if self-driving cars will be so expensive to be niche they won’t actually make a real impact on real people, thus the goalpost of general availability to a consumer.

Re: AI is different

#520

Post by layer8 »

Paradigma11 wrote: post #417
parineum wrote: post #70

Earlier quoted context omitted.

I'd argue that the applications if LLMs are well known but tgat LLMs currently aren't capable of performing those tasks. Everyone wants to replace their tech support with an LLM but they don't want some clever prompter to get it to run arbitrary queries or have it promise refunds. It's not reliable because it's not intelligent.

I think autonomous support agents are just missing the point. LLMs are tools that empower the user. A support agent is very often in a somewhat adversarial position to the customer. You don't want to empower your adversary. LLMs supporting an actual human customer service agent are fine and useful.

How do you prevent your adversary prompt-injecting your LLM when they communicate with it? Or if you prevent any such communication, how can the LLM be useful?
Post reply on HN