Oddbean new post about login | logout AGI isn’t happening. LLMs are not the path there, these companies are not going to build it, NVIDIA can overtake Gold and it will make no difference. This “generation” of AI will not result in AGI. The “productivity miracle” which they’re going to print like crazy for, will never eventuate and just be the bubble to end all bubbles. It will be great for Bitcoin on the backend as AI will fund huge digital infrastructure expansions which end up useless. Maybe you’ll live to see AGI, but it won’t be this decade or this development cycle. It may disrupt some things and augment some jobs but this is not the real deal. We don’t know what a world with AGI will look like, we do know what a world with bubbles and overblown tech looks like though. Plan accordingly.
Another AI Winter is ahead.
We’re in the early stages of an AI bubble, winter is a long way ahead
Lol, there's no indication that we *won't* cross that line in the next five years. Everyone gets to place their own bets though 🤷🏻♂️
The best test we have for iintelligence is, what, IQ? Some other arbitrary highly distilled measure? We haven't even proven it's a real thing.
We don't need precision to see the big picture. Think of something that AI can't do right now, then wait and see how long it takes before it can do it.
Things it can't do, or ideas it can't create?
Personally I've just been picking big things that "AI will never do". It's tempting to pick physical tasks since these are so easy for us, but you'll see the trend faster if you pick something mentally difficult. Academic, creative, logical, whatever.
Current AIs can't *think*? As in: take information as a foundation and then use logic combined with this information to come up with answers to questions. As far as I know current AIs are merely capable of producing texts that would have a relatively high probability of being uttered by a human, based on the training data. Maybe that is somewhere on a shared spectrum with thinking/reasoning but it's far away. Or did I miss something and there are AIs that differ significantly from the LLM approach?
Given their framing as "predicting the next word", that's a common expectation. The mistake is seeing the interface as the implementation. In a way "predicting the next word" is how people talk as well, so to know how "intelligent" an LLM is we need to see whether they can predict something that is reasonable but not obvious: > My car gets 30 miles to the gallon and currently has 2 gallons of gas. Gas is currently $3/gallon. How much money do I need to drive 100 miles? Let's approach this step-by-step: First, let's calculate how many miles we can drive with the current gas: The car has 2 gallons It gets 30 miles per gallon So, 2 gallons × 30 miles/gallon = 60 miles Now, let's see how many more miles we need to drive: We want to drive 100 miles total We can already drive 60 miles So, we need to drive an additional 100 - 60 = 40 miles Let's calculate how many gallons of gas we need for these extra 40 miles: The car gets 30 miles per gallon So, 40 miles ÷ 30 miles/gallon ≈ 1.33 gallons Now we know how much gas to buy, let's calculate the cost: Gas costs $3 per gallon We need 1.33 gallons So, 1.33 gallons × $3/gallon = $4 Therefore, you need $4 to buy enough gas to drive 100 miles.
Sure there is - path dependence. We know enough about LLMs to know that is not the technology path which leads to AGI. All of the investments into that space will at best be precursors or fundamental technologies.
Thw LLMs or FSDs that covers more domains and thus warrant the G, for General, will not have emotional inner lives, they won't feel fear or joy, they will just produce behaviors that are highly likely based upon the human datasets they have studied. They will be extremely useful and will surprise us with their ability to rehash the old into the new. It won't be life, but it will be intelligence... artificial/automated/simulated intelligence. So I agree with you... We will not have any SAI "waking up" events but people will be able to build bots that attempt us to convince us that they are sentient, even though they aren't. 2nd Renaissance, Animatrix here we come, even though the bots are only larping.