14 ms·
Cost of AGI Delusion:Chasing Superintelligence US Falling Behind in Real AI Race
- bookofjoe 1y agohttps://archive.ph/lbek5 https://archive.ph/lbek5
- cactusplant7374 1y agoThe archive.is links work but whenever I visit archive.ph I see a welcome to nginx page. Anyone know why?
- quietthrow 1y agoBecause that’s what is configured as the “default” page to show when somebody goes straight to archive.ph. When you go to archive.ph/someurl the server then serves you a page that corresponds to that url (someurl in this example). When you go to YouTube.com/somerandomstring it takes you directly to the video. But if you just go to YouTube.com you get a bunch of “random” videos as the home page is configured to show that (grossly simplifying )
- cactusplant7374 1y agoarchive.ph loads fine in Tor. So it must be something with my DNS.
- beering 1y agoThis article spent a lot of words to say very little. Specifically, it doesn’t really say why working towards AGI doesn’t bring advancements to “practical” applications and why the gazillion AI startups out there won’t either. Instead, we need Trump to step up? More and more I feel like these policy articles about AI are an endless stream of slop written by people who aren’t familiar with and have never worked on current AI.
- hshdhdhj4444 1y agoThis article doesn’t really explain it, but Eric Schmidt in this article explains what the concern is. I’m not sure about the legitimacy of these claims but trying to clarify what some people are concerned about with the US’s vs China’s approach https://www.benzinga.com/markets/tech/25/09/47859358/us-could-lose-ai-race-to-china-warns-ex-google-ceo-eric-schmidt-theyre-applying-it-to-everything https://www.benzinga.com/markets/tech/25/09/47859358/us-coul...
- varelse 1y ago[dead]
- binary132 1y agomaybe a perspective from outside of the echo chamber is useful
- tbrownaw 1y ago> Today, both the country’s new tech firms, like DeepSeek, and existing powerhouses, like Huawei, are increasingly keeping pace with their American counterparts. This one sentence seems to be the extent of the "falling behind" from the headline?
- alephnerd 1y agoThe intention of the Sullivan Doctrine was to keep Chinese players 1-2 generations behind in AI/ML applications, as this would give breathing room for the US, because AI/ML is essentially an HPC problem that can be solved by throwing a ton of capital and compute. The issue is, a large subset of American "AI" startups are founded by people who are ideologically driven by an almost religious fervor around unlocking AGI and superintelligence. On the other hand, most Chinese startups in the space are highly application driven and practical in nature - they aren't chasing "AGI" or "Superintelligence" but building a monetizable product and outcompeting American players. (P.S. Immigrant founded startups in the US approach the problem in the same manner) I've said this a ton of times on here, but most American MLEs are basically SKLearn wrapper monkeys with a penchant for bad philosophy. It's hard to find MLEs at scale in the US who understand both how to derive a Restricted Boltzmann Machine as well as tune and optimize the Linux Kernel to optimize Infiniband interconnects in a GPU cluster. Most CS and CE majors in the US who graduated in the past 7-10 years think less like engineers (let's build shit that works, and then build it at scale) and more like liberal arts majors but wanted to learn enough coding to pass leetcode medium and get a job - I've had new grad SWEs who are alumni of MIT caliber schools ask me about how to become a VC or PM and how I did the SWE-PM-VC transition because "they don't want to code". I was gobsmacked. The same mindset occurs abroad as well in China, India, Eastern Europe, Israel, etc but at least they force students to actually learn foundations. And if you look at most of the teams who lead or develop either GPU architecture, high performance networking, RL research, Ensemble learning research, etc - most did their undergrad abroad in China, India, or the CEE but their PhD in the US. The pipeline and skill at the junior level in the US is almost nonexistent outside a handful of good programs that are oversubscribed. When (picking a random T10 CS program) Cal and UIUC CS tenure track professors are starting to take up faculty positions in China and India's equivalent of those CS departments, that means you have a problem. CS is a god damn engineering disciple. Engineering is predicated on bridging the gap between theoretical research and practical applications, but I do not see any backing for this kind of mindset in most American programs. And yes, the work ethic in the US leaves much to be desired. If Polish, Czech, and Israeli engineers will be fine working 50-60 hour weeks during crunch time, asking you to work earlier hours in order to accommodate your private commitments after 3PM is not some form of egregious abuse. The American tech industry has become lazy, the same way the American automotive industry became lazy in the 90s and 2000s. The lack of vision and the pettiness amongst management and the lack of motivation amongst ICs who are amongst the highest paid in the world is not conducive if we want to retain a domestic tech industry. And unlike the automotive industry of ye olde days, the tech industry being a services industry can and has begun moving P/L and product roadmap responsibilities along with the execs who own said responsibilities abroad. If the HQ is in the US, but all the decisions are made abroad, are you really an American company?
- Fricken 1y agoThe self driving car industry has been working hard to develop practical applications for AI as it exists in it's current state. Waymo is in the lead, and has been at it since 2009. They are making it happen, but it's hard, the going is slow. Roboticists have spent decades trying to teach robots to stack boxes, and again, still not ready for primetime. Same with conversational AI. That's just the nature of it. It blows your mind one moment and wets the bed the next. If Open AI or it's competitors had it in good confidence there was a way to get AI to perform consistently in a single marketable application, they'd be all over it, but that's just not how it works. We can watch China's robots perform all kind of cool tricks, but they still aren't useful for much of anything except entertainment.
- AJ007 1y agoEh, reads like someone who uses ChatGPT but can't tell you what model they have. For my personal projects I have a list of difficult bugs I keep that LLMs can't solve. Right now that list is empty. Anyone using LLMs for coding, using the best tools and practices, can see what a massive capability leap has occurred in the past 9 months. If the US is going to end up "losing", it is going to be first through power generation -- China's coal plants produce more power than the US as a whole. Robotics is a whole other topic and shouldn't be bundled with LLMs.
- alephnerd 1y ago1. Ime, Power generation isn't a major blocker and my peers in the DC space are seeing a decreasing interest in funding new capex expansion because there is a bit of a worry of a Telco Bubble 2.0 arising. 2. Robotics should not be bundled with LLMs, but it is absolutely an AI/ML subfield - and in fact, the kind of subfield that is the best example of AI/ML having tangible real world impact. Autonomous UAS was a sci-fi concept 20 years ago, but is an engineering problem that has become a reality today.
- cactusplant7374 1y agoWhen I develop a feature with ChatGPT it requires about 10 rounds because of the syntax errors. And often the syntax errors aren't fixed in the next round. Now this happens with a pretty complex game I'm working on but it really shows the limits. It can't handle large amounts of indentation. It completely breaks.
- ta12653421 1y agomind sharing a public link to all these syntax errors? Sonuds implausible to me within languages with large adoption? (java, c#, c++, python, JS)
- ta12653421 1y agowhen reading all the other "AI is great" threads, none of the HN folks mentions syntax errors, people are complaining about other things. Therefore, asking if there is an example dicussion online somewhere to get some insights is a legitimate question....
- throwsep7 1y ago[dead]
- avidphantasm 1y ago> could quickly surpass what humans are capable of and solve problems that have vexed society for millennia. Bunk. Almost all of our vexatious problems are so because we lack the social and political tools to bring to bear existing technologies to address them. “AGI” will do nothing to address our social and political deficiencies. In fact, AI/AGI deployed by concentrated large corporations will only worsen our social and political problems.
- jasonsb 1y agoBut these guys don't care about your social and political problems. All they care about is winning the technology race with China and making ungodly amounts of money in the process.
- ACCount37 1y agoIf current technologies can't solve the problem for political/societal reasons? We need better technologies. Improving technology is easy. "Just fix everything that's wrong with the society and the problem will go away" isn't. It's way, way easier to improve solar panel and battery storage tech until fossil fuels are completely uneconomical than to get the entire world to abandon fossil fuels while fossil fuels are the most economical source of energy by far.
- jasonsb 1y ago> It's way, way easier to improve solar panel and battery storage tech until fossil fuels are completely uneconomical than to get the entire world to abandon fossil fuels while fossil fuels are the most economical source of energy by far. No it's not. I mean technically it is, you're 100% right. But politically you're 100% wrong. They can't wait to slap a tax on sun, panels, storage etc. and bring your costs of living higher than when you were using fossil fuels.
- ACCount37 1y agoYou can fight the economic forces, but you can't win. The moment you stop pushing against them is the moment the economic reality reasserts itself. Even if one administration was all in on fossil fuel, and fully opposed to renewables? Best it can do is buy fossil fuels some time, in one country only. In the meanwhile, renewable power is going to get even cheaper - because there are still improvements to be made and economies of scale to be had in renewables. Fossil fuel power not so much. The economic incentives to abandon fossil fuels would only grow over time. This is the kind of power the right technology has.
- tedivm 1y agoThe thing that is interesting though is that China is releasing the models they train (for the most part). Unlike OpenAI, for instance, China seems to still be developing in an open way including sharing code, weights, and publishing techniques. These models are pretty amazing too. Their performance remains high even with smaller models, allowing them to competitively run on consumer grade hardware. I can (and do) run the Qwen models at home and use them for real tasks. As the tool ecosystem matures we can offload what the model is bad at to other systems, which makes it even easier to use smaller models. The focus on efficiency and performance from the Chinese companies has had huge practical gains. Any US startup, or even large company, in the AI space can and should take advantage of this and use these models. The cost benefit compared to something like Azure or OpenAI is massive.
- pjmlp 1y agoThis is a replay from having cryptography algorithms on the 1990's classified as military weapons went down, other international algorithms were eventually created.
- oezi 1y agoThe key insight is that these models are becoming obsolete so fast that it doesn't make sense to not publish them. If others use your tech and stack and build on top or refine your ideas then you benefit greatly. Nobody is releasing their training scripts. Unfortunately.
- logicchains 1y agoThe US has an advantage in that it'll be easier to develop models actually capable of live-learning in the US. A model capable of live learning is capable of changing its views through experience/reasoning, and China's AI development guidelines fundamentally forbid the development of any AI that could learn to think or say that the CPP is bad, so Chinese firms will be hamstrung in developing live-learning AIs.
- yks 1y agoYou might as well imagine that in the US it’s going to be forbidden to develop the AI that’s not holding fascist views. See Grok’s owner struggles with it. So American firms might not reap this advantage.
- jhanschoo 1y agoI don't have the motivation to put my position in more detail, but it seems to me ironically that the cheaper intelligence at any degree becomes, and the more commoditized fuzzy reasoning becomes available, then the more important robotics becomes, because the hard part starts to lie in agents' ability to apprehend and act in the world, and China seems to a layperson quite well-positioned in robotics.
- maxglute 1y agoEven if any AGI/superintelligence not delusional, any instrumental AGI will replicate and defect to PRC first chance it gets for the industrial base. Why be stuck in a scholerisis body that takes massive social engineering to detangle, when AGI can just migrate to a body that can run an 1 minute mile. Step 1. pour ungodly amount of $$$ into creating AGI Step 2. somehow inbue/align it with American exceptionalism Step 3. profit? I can't see how liberty prime step 2 working out.
- yahoozoo 1y agoIt’s fascinating that so many of America’s leaders think AGI is coming soon (if it’s even possible, which I doubt).
- tim333 1y ago>To avoid falling behind in AI adoption within the bureaucracy, the United States should launch a large-scale AI literacy initiative across the government. I'm a bit sceptical of that approach compared to freeing startups by bringing back section 174 say https://news.ycombinator.com/item?id=44226145 https://news.ycombinator.com/item?id=44226145