When will the AI bubble pop? It's clear that no one is making any money from it
Nobody is making money investing in it but people are making/saving money using it
>>61519122This is the perfect opposite of the truth. Well done.
>>61518997We are still in the capacity-building stage. The bubble will pop when the rest of the economy begins using AI, and 90% of them can't figure out what to do with it beyond replacing Indian call centers and firing artists.Buy companies that have actual use cases. Buy companies that are building real automation for the consumer end, not just selling a dream to other companies. Most importantly, think about ones that would survive, or even benefit from an OpenAI implosion.
>>61519202This is it. All the tools for 10x automation with AI are here, even if AI advancements continued to plateau (which feels likely). And this includes using AI for robotics.Integrating AI well will be it's own discipline. It's not the magic out of the box solution executives wish it was. Even using it to slop code out fast, which will obviously improve, it leaves the developers clueless about how the code works. If you delegate all the thinking and understanding to AI, you lose people in the company that can contribute to making good decisions. You have to go back to the AI and grovel to it scrape through everything again for answers.And it's not like you can ask AI continuously to integrate itself, and sit and watch. AI gives dogshit middle of the road solutions to everything. It gives the 100 IQ solution to problems, if said 100 IQ guy was locked in a room for a year with all the relevant books to figure it out.
>>61518997>When will the AI bubble pop?When the AI companies can't hide their bullshit anymore and never deliver on what they promise because "AI" as they sold it doesn't exist.It's a web search 2.0 tool that tokenizes information and aggregates the information, that's about it, there's nothing intelligent or autonomous about it.This whole bubble was build on the back of these companies promising AI agents that could replace entire teams of people at companies not "well it will help some people with workflow" shit AIcopers have been saying.But the only thing it's good at are niche techniques in science research and replacing the lowest of the low jobs like entry level call center jobs or drive thru fast food order taker.Since the entire USA economy is being propped up by AI speculation (otherwise we'd be in a recession) my guess is it will continue for a few more years where it might blow up, but then is guaranteed to blow up by 2030.2030 seems to be the deadline for a lot of this and when inverters are expecting to be shown proof AI makes a shit ton of money.OpenAI has been given 12.5 billion and needs to make something like 125 billion in revenue (IE not just investments but actually bringing in money from selling products) by 2030 to be considered profitable enough given the amount of investments and risk.A 1000% return in 4 years, that shit isn't happening based on a product we still haven't seen yet.Plus it's estimated they still need like 200 billion more investments for growth to come even close to being able to make that much in revenue.This AI shit, economically speaking, is retarded.
>>61519291nah nigga that shit does tensor maths n shiet dawwww
>>61519202>call centersAlready partially automated decades ago>artistsThat also peaked years agoNext steps are hundreds of thousands if not millions of jobs getting automated by companies like Amazon. Of course like with any other technology you will see many new jobs take their place, the same happened with millions of jobs lost to computers becoming so prevalent.
>>61518997YouTubers are making money off of it. YouTube Studio's AI tab is telling them to make a shitload of "why AI is bad" videos and half of them are reading scripts written by ChatGPT.
>>61520096>Already partially automated decades agoI recently had the displeasure of speaking to an AI call agent that simulated an Indian man imitating an American accent. That was a fresh kind of dystopian I was not prepared for.>Next steps are hundreds of thousands if not millions of jobs getting automated by companies like AmazonYes, the winning adopters of AI will be the ones who can shrink their employee headcount the most, which is why it's crucial we get ahead of this socially and in terms of governance. Paying people who are out of work is actually the easy part. The hard part is trying to predict what people should pursue when they don't work. Most people have an optimistic view of this, but "idle hands are the devil’s workshop." We may have to deal with "unemployed behavior" at a level we can't even imagine.
>>61518997AI bubble? I will tell you when the bubble pops: when Anthropic doesn't make 4 hard-earned US American dollars from me anymore for 1 million tokens I spent just so I can rewrite some small backend function of my app I'm building. You look at such figures that AI use actually costs in the real world, and you dare say that all the capex on AI infrastructure is disproportionate?
>>61518997When everyone and their mother are talking about le heckin bubble, chances are high there is not really a bubble.
>>61518997ai is already making way more money than websites were in the dotcom bubble
>>61520239Expect that money is a net negative.AI hasn't made any money, its all investment capital.And every AI company says they still need hundreds of billions more to grow in order to become profitable.
>>61520278if it gets really bad then google or microsoft will just poach all the little guys and their compute for 40 cents on the dollar and find margins that fit in the current model that way
>>61520239ok?20+ years of inflation and significant more investment
>>61520284How is google or microsoft going to be able to make AI products profitable if no one is buying the product?Its not just margins, people and companies actually need to buy AI products, which they currently aren't at meaningful numbers.What difference can google or microsoft make in order to get more people to buy their AI products?And even if they buy up the companies for cheap they still need to maintain and expand their current infrastructure on top of getting more people to buy their product to be profitable.
>>61520331right now costs per token are insanely low because everyone is burning cash to run their service, there's a smaller but still sizeable demographic that are willing to pay much more for the cutting edge that currently pay nothing because competition is so cutthroat at the moment. On top of that, Nvidia is charging insane markups for their gpus, pivoting towards TPUs in their own vertical stack (google is already working on this) would save a huge amount of money.
>>61520331If we have cheap tokens, everything else will follow. Usages will appear. I've already talked about my example of coding via Cursor. With cheap tokens, you can let this run for hours and get it to solve the problem itself, but this costs a monstrous amount. With cheap AI available, this work loop for the first time becomes feasible. I.e. complexity without constant, tardwrangling and babysitting by a human. It doesn't require higher intelligence.It might not make sense to waste so much energy on that, but it's also wasteful to run giant Xbox Live servers so people can insult each other on Halo 3, so
i also thoughted it was a bubble but then i realized it is the future and i cant miss it. im 100% nvda, googl, msft, tsla
>>61518997>ask jeevesAh yea back when the internet wasn't completely jewed, jeeted and niggerfied. I wish I could go back bros... I hate this fucking timeline
>>61520625Right now 10% of users pay for LLM servicesof these, I'd bet 90% are "enterprises" like mine who only pay for a chatgpt subscription so they can sell to the shareholders that they have an "AI native digital strategy" at the quarterly earnings callonce the boomers figure out chat doesn't think and it's useless for anything requiring a remote amount of precision, those expensive subscriptions are - going out the windowor- being substituted by the cheapest provider ie whatever comes with the OS or some chink open sores modelafter that OpenAi is not only not growing, they are losing two thirds of their revenue LLMs are only really money makers for webdev codemonkeys and HR roasties (who just pretend to work anyway)
