Not all professionals require enhanced processors. Also lots of people want better screens and more ports. Don’t let your assumptions limit you imagination.
I had the opposite take on LLMs using Kahnemann's framework.
Maybe I need to reread his book, but I think of System 1 as "fast, reflexive, instinctive" mental activity, and System 2 as "slow, deliberative, rational, verbal" thought.
So exactly inverse to this article, I characterize LLMs as 100% System 2, and would say that what's missing in AI is the more "embodied" System 1 intelligence.
My current thinking is that the experience of consciousness is closely tied to that embodied, emotional side of our nervous system, and is 100% lacking in LLMs. Anthropic people will say "the models have neural correlates of emotion", and that's fine, our cortex also has neural correlates for emotions and for much of what happens in the cerebellum and brain stem. That doesn't mean that that's where the primary function is implemented.
> So exactly inverse to this article, I characterize LLMs as 100% System 2, and would say that what's missing in AI is the more "embodied" System 1 intelligence.
I've read Kahnemann's book and how I've seen System 1 / 2 typically used in the context LLMs often makes them seem like poor fits to what's being referenced. The concept was created to broadly group types of human reasoning but it never struck me as rigorously defined enough to be a technically precise distinction. Applying what was a mildly useful directional thumbnail referencing human thinking styles to LLMs, which can mimic human output but via completely different processes, quickly goes off the rails. As we get more detailed and precise, the System 1 / 2 distinction seems to increasingly diverge from how LLMs behave.
A quick example is how basic Flash-type chat models without 'extended thinking' or MoEs will keep suggesting different solutions instead of stopping after a few failures to step back, gather more context and diagnostic data and evaluate changing approaches or questioning the question. When I do that, the 'interrupt' is often a 'System 1-ish' intuitive feeling like "I suspect I'm barking up the wrong tree" then, once stopped, I'll switch to System 2 to analyze. But in LLMs that "interrupt" behavior typically arises from adding more System 2-ish thinking. N=1 but that's not how I usually 'get there' and is orthogonal to what the System 1 / 2 model suggests.
>Maybe I need to reread his book, but I think of System 1 as "fast, reflexive, instinctive" mental activity, and System 2 as "slow, deliberative, rational, verbal" thought.
Current LLMs are the opposite of slow and deliberative, though? With the current transformer architecture, it seems like its function is more or less "look at this text and make a guess which word comes next", which is very much a fast and intuitive task. Any reasoning comes from CoT, which is more analogous to System 2 thinking. But that System 2 thinking can't function at all without relying on the representations from System 1 thinking (which are formed in pretraining).
So you could argue System 2 thinking is some part of the LLM, but it can't be 100%.
And when you think, how do you know which word will appear next in your mind? Or does your system 2 thinking seem to emerge as a spontaneous stream of words/thoughts?
I actually think that AI has little to do with the brain / human intelligence, but in this regard, human thinking is also spontaneous, where the next thought just seems to emerge from the combined effect of all inputs to your brain (including the memory of recent thoughts).
System 2 is slow because it is sequential. Every output depends on all
the prior outputs. Like a transformer. (Even if this comes about as the result of massively parallel processes, both in the brain and in the model.)
System 1 is more of a parallel thing, operating on the output as a whole rather than in sequence. Much more like diffusion.
Yeah the framing seems inverted. We have a faster system 2 with llms, and system 1 is what decision models are hoping to be, but so far they just seem to be system 2 models without language output.
The terms don't seem to translate well, possibly because they were the wrong framing to begin with. Possibly also because the billions of years of progressively complex evolutionary constraint that brought about animal cognition does not seem likely to be reproduced by running backprop on human language artifacts.
> the people at Whole Foods who have to people find their “prime code” in people’s apps because it’s impossible to find.
This used to be easy and then they moved it somewhere counterintuitive in the app that my brain refuses to remember. I swear.
Now they’ve put a QR code on the checkout console, and scanning that opens the Amazon app to the right spot. Scan a code to scan a code… modern life.
(WHat’s really ridiculous is that I’m paying with my Amazon Prime VISA card—why doesn’t that automatically get me my prime discounts? It’s my default method of payment on the Amazon site, it’s not like they don’t know about it. No, I need to jump through some other hoops to “enroll” or “link” my card or something for that to work. eyeroll )
Prime Visa: that’s the one that gets me. Same for me. I have restaurants I go to that have loyalty programs that can figure me out by my card. But I got the card from you, and to go to the same store all the time, and you can’t figure it out? Come on.
But they could also make the app way easier. Last, I checked the stores didn’t move around randomly. They’re always in fixed locations. So why not ask for location permission and just automatically pop the code up or make it a simple option at the top of the screen whenever it detects you’re within 50 feet of a Whole Foods?
All I can think of is that they advertise the whole thing to get people in for “savings“ and then make it as hard to use as possible, so no one actually uses them.
I hope you have been installing security updates for the OS version you have. At some point, those end, and then you're looking at a different trade-off in not upgrading.
I think they were asking, "If this is a model produced by Meta with an API designed to be compatible with OpenAI models, but it's not actually an OpenAI model, why is Meta hosting it on Azure?"
I guess at peak times they might need more to the extent they offload to azure and still sell excess off peak, or some announced partners required all data stay on azure?
In 2003, Gates was no longer CEO; I think he was President of the Board and something like "Special Advisor"? None of the people involved reported to him, so he may actually correctly have felt he could not tell them what to do?--so he stuck to venting his frustration.
Exactly. bluegatty has great points for a CEO role but Bill stopped being CEO in 2000 three years earlier. The board and advisors cannot direct the company internally - this is by law. Separation of oversight and management. They can tell the CEO what to do but the CEO decides. Then they can fire the ceo if they don’t like it.
If the President of the board has an issue with how the company is fundamentally operating, they go to the CEO. If the CEO doesn’t understand or doesn’t prioritize the concerns of the board, then they get replaced.
He’s may not have been the CEO… but he was the boss of the CEO and represented all the individual owners of the company.
True but the separation of concerns is to put a wall up. Billg was relating his poor experience to the company he oversees. He is strictly forbidden from micro managing at this level. Eg "this is a problem" vs "fix the problem this way". Not his job
Even if he were still CEO, a message of, "This sucks", he does not have to explicitly add, "Fix it, please". That is well implied by the hierarchy. Your boss notes a problem, underlings try to correct.
I can see why you might think this but I think it is incorrect. Particularly because he's not in the hierarchy. Billg is not the boss at this point and has no meaningful effect on one's job unless you are in the top dozen execs and even then this is just an email about a user experience, it's not a command. Ballmer is your boss and the old man yelling at the cloud can keep screaming into the wind. Point: did anything change from this email? I don't think so.
It is helpful for me to understand corporate dynamics and how they actually work (or as supposed to) instead of a simplified version from outside. My brother is a CEO of a big company so I get to hear it from his side. Doesn't mean I'm right, I'm not the CEO.
[edit: softened the language]
[another edit]
Actually he isn't the CEO anymore. Private equity purchased the company years ago and he was made CEO from outside but in the industry. Eventually the PE people wanted to cut everything to make the company look good for a purchaser so that means high earners in the company have to go (this is a highly technical company with PhDs and researchers, etc).
But guess what? Those are the people that make the company special among other things. They demanded he fire some top people. He wouldn't because it would destroy the company's value. So they fired him and put in one of their lackeys to do the slaughter. They kept him on in a different role that I won't say because I don't want to reveal too much. Flash forward the company has not sold and is likely to have lost 50% of its value. I admire him but since they were going to be fired anyway shouldn't he have sucked it up and got a better price for the company? The PE people were total jackasses so working for them another day was poison in his veins.
I’m not talking about micromanaging, I’m talking about relaying the information to the actual CEO (Steve Balmer I believe?) and ensuring they achieve the goals.