A thread for unboxing AI

Managing the process of mass-AI checkups. At the end with , say, the same amount of radiologists we have today they will "produce" 10, 20, 50x the output.

And then fewer people will go in radiology, or will go there to be the finetuner of AI processing within radiology but with a different education to achieve that and so on.

Now apply the same to basically everything AI can be used for at scale, that's just ultra-massive increase in productivity, everyone is better off gigantically, some more than others.

So the value will be coming from the AI and there may be a few 'highly trained experts' doing some admin. For a while.

Yes apply to everything (includign admin). I agree we will be massively better off. Just need to distribute it.

It's almost like your arguing for reasons to not go fully commie here. As someone who's mostly pro cap, AI is probably your best gift in allowing for a us to ditch capitalism to w/e degree and adopt socialism. It's literally a gateway to afford the economic feasibility for governments to pay out large scale social programs - something that the naysayers either can't articulate or don't care enough to do so without it.

The issue is adoption, and while that may be disagreeable in term so timeframe, it at least becomes a luxury to discuss when the mechanism can actually allow for people to live a more fullfilling life during that time.

We have already headed in that direction with better working conditions, 4/day weeks, better govt support and so forth. That progression isn't directly tied to either capitalism or socialism, its tied to good old fashioned innovation, and I think that's the who ****ing ballgame - and AI is just that.

I think were starting to understand each other.

I'm not anti-capitalism. It had strengths and weaknesses but combined with socialised services it was probably as good as we could do. I'm not into the capitalism vs communism debate. I'm saying capitalism as we have known it is coming to an end whether we like it or not. Humanity is burning through it like a star burning through it's hydrogen - arguing a preference for the hydrogen stage isn't going to make a jot of difference.

I'm pro innovation and technology. The future will be awesome and I think the section of the left who are anti-growth are totally misguided. But AI/etc is not just innovation just as agriculture was not just innovation. These are transformative changes and AI/etc is going to make the shift to agriculture look like a walk in the park.

The bit I want to preserve is democracy - not least for moral reasons. Otherwise I would probably be looking at full blown communism (the innovation embracing kind) as the best form of authoratarianism. Although I'm unconvinced that all authoratarianism wont end up pretty much the same

The problem chezlaw which you don't address is who decides what gets produced and which investments happen in your model.

That's what you don't address basically at all.

You mentioned you want ownership of the AI to be public. That presumably means a public committee of public employees decides everything.

That means making every important sector of the economy THE DMV

I mostly agree about the scope and inevitablility of the changes that are coming. I am less certain that humanity will be "massively better off." We will live longer. We will feed people more efficiently. We will house people more efficiently. But part of me thinks that we are a crude form of life that needs to have goals and sometimes achieve those goals in order to avoid crushing listlessness and loneliness.

This is all part of my long-running concern that our technology will change much more quickly than our ability to adapt to our technology.

The only problem I have with that post is it sounds like it's a disagreement when it's where i started the conversation in SMP a few decades ago. I'd add - and they will vote! Careful or I'll come up with a classic chezlawism like 'we're better off being worse off'.

This should imo be the stuff of real politics but we're too busy shouting at people while denying the profound change that is underway.

Apart from the politics it's a fascinating topic which includes a solution to the problem of evil.

Why would goals cease to exist with more material prosperity?

I don't want to do or say anything that will elicit the bolded.

Even worse when they start to make sense

Don't worry, Skynet will have launched the nukes by the time that happens.

Dont worrry, I didn't mean you.

Yeah, still not in any danger of making sense, I see. Prosecution rests.

There was never a case. You're far too well protected

Protected? Are there tariffs on my competition?

It's hard to set goals, much less achieve them, when there are virtually no consequences to achieving or failing to achieve your goals.

Also, as a world, we are not trending in the direction of using our free time in healthy ways.

I was in an abnormal psych class and we aomehow determined, or was determined that the one word secret to happiness was progress. It made sense when you look at spoiled millionaire winners or cancer survivers and such. I remember strongly wantinf to live my life along that where i could - and still do, and realizing how that worked for the greater good in my life.

Who knows how much of the drive or opportunity will be there but i agree that humans need things to build and work towards, in multiple ways.

with actual abundance/material prosperity they can get back to what was the source of meaning for like... most of our history for most people, having children and grandchildren (if you were lucky to live long enough to have some).

do you realize humans lived for all of their history except the last 2-3 centuries in super stagnating societies where nothing changed for the good during a single lifetime and change, when it happened, was almost always horrific (wars/diseases/natural catastrophes)?

even when a lot of the world was "unknown" (undiscovered on the map), 99.x% of people never moved more than 20-30 miles from the place they were born in.

Yes, and not all that science fiction agrees with you.

If you didn't control for age, then I suspect that the answer would not favor your position. I wasn't talking solely about about depression, but depression and feelings of uselessness are quite common among the elderly.

I haven't seen any studies that compare the psychological health of trust fund kids as opposed to a control group so I can't answer that question. In any case, even if you are a trust fund kid, I suspect that it is quite a bit easier now to talk yourself into believing that you are setting and achieving goals than it will be 100 years from now.

I am not claiming to be certain about any of this. Anyone who claims to be certain about exactly what the future will look like is almost certainly a fool.

There is far more abundance and material prosperity in the Western world than there was 100 years ago. But that increased abundance hasn't resulted in people finding more meaning in their children and grandchildren than they did in the past.

If i'm picking up what you're putting down here than you're mostly agreeing with me.

I don't think a workforce is necessary for human beings to satisfy their drive to self growth and "progress" and I don't think that having the opportunity to better provide for those financial needs would eliminate or kill that desire - it's too multifaceted obv.

I think that in the general macro sense of it all, something that roc was kind of eluding to, i think, is that the transition from away from human beings providing/controlling w/e you want to call it to being forced fed like cattle in a sense from AI, isn't going to improve the human psych in a general sense and may through things a little off balance if it happens too quickly.

But im really tired and cant articulate anything right now so maybe explain later.

This is pretty much the crux of the issue. I spent some of my career in a field where errors get people killed. Not killed in the sense of a miscalculation in an excel sheet meaning some guy you don't know dying 10 years from now, but killed as in you might lose a colleague in the near future. Out of curiosity, we did a comparison of the various AI models, ranging from simple ones to the most sophisticated ones. Out of the bunch, the simple ones were the best, because they had such hilariously bad answers that pretty much anyone would realize you should never use any of it. The advanced ones were much worse, because their errors were much more difficult to spot, and the were also better at using "industry rhetoric".

A big problem with AI is that a lot people who use them do not realize what they are. They see their output as similar to that of a calculator, database or search engine, because this is the digital tools they grew up using.

However, most AIs are actually pretty ****ty calculators because their rules are flexible, they're poor databases because their relations between information is not fixed and they're bad search engines because they are based on aggregating information, not showing you the instances.

That doesn't mean they are bad at what they can do, it just means it is important to understand what it is they actually do and its limitations.

Keep in mind you were testing generalized LLM for very specific life or death tasks. AI isn't LLM only, and especially it isn't general LLM only. And even with general LLM the specialists in a random sector aren't necessarily specialists in prompting LLMs...

You can have tailored AI for specific purposes tested very aggressively for a long time before being deployed, it just delays the results a bit, but at the end they will be better than humans at those taks.

In some very real sense we already have accepted that for planes and trains for example.

And even if a human operator can still be needed , or we want to have one because of risk aversion or whatever, AI doing a growing portion of the job will allow higher productivity.

When hearing people critcise AI, you can normally replace 'AI' with 'human'. It's not replacing the calculator or the database, it's replacing the human who used calculators and databases.

As with humans, it's harder to spot the more advanced idiot with the jargon than the complete idiot. Although when TD says "Out of the bunch, the simple ones were the best, because they had such hilariously bad answers that pretty much anyone would realize you should never use any of it." I was assuming that made it a general.