Coping on the big Albert I
You saw the title. I'll get to the point:
Mindset is EVERYTHING - your belief system, especially in the modern age with all these self-tailoring recommendation algos, will define not only you but your environment. You will live in this belief system, conflicted or not. With that being said, I'd say the optimal spot to take on this topic (at least for me) is one of healthy suspicion. Don't believe every single claim that AI is monster big brain atp and going to take over the world, and don't believe they're just mechanistic parrots - hard surprisals should be backed by hard empirics. I also personally think catering towards the AI-optimism pov is not bad either - just from the general rule that you'll probably like your life a lot more if you think consistently optimistically in general. In the end, no one really knows what'll happen or how this will all play out - that's just the uncertainty of the future, and we all have to live with it - so it's best to look in favor of the future, rather than shun it.
I like this analogy: we use cranes to lift weights, but that doesn't mean you shouldn't go to the gym anymore. There are things we do simply to do them, for benefits which we hold as like implied, universal almost. Losing them would almost be losing part of our identity; our health; our lifestyle. In the same way, we shouldn't lose any sort of mental "exercise" in favor of just using AI. You just shouldn't be using it when you really want to get the stuff done. I'd personally separate work into 3 classes: 1) you really want to get it done - just fully use AI, 2) you want to get it done but also learn from it - somewhere in the middle here, kind of use AI maybe as a verifier or something, a lot of this is situation specific anyways, 3) you're actually doing this to learn/"mentally exercise" - no AI of any sort. I think this is a good approach in general when dealing with AI for work. Additionally, some of the "exercise" we do is so near and dear to us that we exalt the top performers of the orthodox traditional; even though cars have become the main way to travel atp, that hasn't stopped us from doing marathons, from cheering on the GOATs like Eliud Kipchoge and Sabastian Sawe. These are canonically called sports I heard, and I heard the cream of the crop pays well and are very respectable in society - I don't think AI will be affecting these spots of isolated tradition at all, actually. Simply because they define the peak of human - yes, human - feat. There will be no Lebron James AI - or at least I don't think he'll be named the GOAT. This is not restricted to what's generally seen as "physical" work either. Mental work too - like chess. Whether you were born with it (talent), or you changed your identity to fit it (hard work/practicing), or something in between, to be the best at these sports is valued and respected in society; again, they define the peak of human feat; something held so crucially to the identity of human society. Heck, even "social" work - the ability of communicating with others - has a very innate sense of identity tied to it too. You introduce yourself to someone - reveal a part of your identity, however superficial you want - when you want to have a conversation with them, after all. I believe these aspects will still all be valued even with AI at present, and in the future. And lastly, AI doing more work just means more time enjoying ourselves, touching grass, spending time with others, SOCIALISING, and just having fun! Experiencing the Life^{TM} whatever that means lmao.
As of right now at least, there are many things AI still can't do, although at the pace its going it might be able to do many of them in the future. But for someone living purely in the present, that's not much of a problem as the money and respect is still running where it needs to run presently. Thus, it's not a doomer situation unless u really treat it as such - as you think that the future is cooked and u only think about the future. Moreover, AI simply does not think the same way we do. It is not human. It was not born with the neuron-weights we were. It uses a silicon brain. An architecture of 1's and 0's along with a bunch of matrix multiplications. In the same way, even if AI can generalize to do some task, it will probably not do it the same way we humans do it, which can lead to downsides and upsides on either method. Thus, I believe its optimal to keep both sides in play and use the upsides of both. Let's say you're running away from someone in a very urban city (hopefully not in reality) - you wouldn't want to drive the entire way around the bustling streets; you'd maybe drive 10 miles around to get some raw distance in, but then ditch the car and start running and hiding in buildings - you wouldn't want to drive your car through a building, of course. And of course, then there's the verifiability part. It's just always good to have two pairs of expert eyes on something to verify it's true - AI AND human experts both having roles is just an optimal plus altogether in many cases; better than just AI or just human experts. Lastly, there's Moravec's Paradox - there's a possibility that AI simply won't get to the stage of doing certain tasks we can do in the manner and quality we do them, simply because evolution trained us for so much more time than AI did. In any regards, we should keep at the present tasks which matter regardless.
I want to go over 2 last points, amalgamated into this last 4th point. The first ties into the 2nd, and it's on identity. There's a post-discursive and pre-discursive ideology on who we are and what we do - i.e., fancy talk for there are views on things which are un-superficial; sentimental. This happens to all of us, I believe. This identity shapes who we are, as an individual and as humans, and losing it would be losing part of our identity. In that sense, we gatekeep them and keep them in respectable and meaningful places regardless of if they're replace-able post-discursively in a sense. These we should not lose; they define/defined/hopefully will define us. And of course, lastly, especially for technical roles, we discussed verification but there's also a level of creativity involved with a lot of these fields. A lot of stuff, e.g. ideas in pure math or some other field, not measurable in definitive quantity, other than maybe in practical impact, but definitely not purely, purely for the sake of it. Basically, I'm saying is you reach a certain technical threshold and then its all on creativity and maybe a touch of randomness. It's these skills which can always benefit from an extra different perspective, or at least almost always. So despite having AI research these areas just as humans once did, humans can STILL research these areas and STILL be actively contributing, and not just a little, given that these stuff are not really even measurable so "little" is quite a subjective term once you reach past that threshold. I think then that this monotonically positive composition which holds in both these aspects of creativity and verification (and possibly more aspects I haven't thought about) are good arguments against the so-called AI doomer perspective (i.e., AI is going to take all of humanity's jobs in the future and we'll be doomed). idk tm tho. just my take. ight peace.
P.S. Oh and last thing, things made in the present will still matter in the present. The first person to do this, the first person to invent that... they all make their respectable and hopefully-well-paid contributions, achieving greatness despite maybe having multiple duplicates or replicatables/replacements in the future. In that sense, innovation in the purest sense will still have value, since after all, we live currently in the present.
P.P.S., there's also just hidden information in general, like the krabby patty secret formula. In this sense, you won't be able to replicate that using AI, since well... it's secret (although innovators in the know may try to use it, but again that's a task/work in and of itself too, and i'd say very very often requires that same verification/creativity/pass-the-threshold-and-its-net-positive-if-you-and-AI-both-take-the-role aspects to that work as well, or maybe AI's implementation in the system is seen as too big of a risk, despite the possible gains if any at all. lifes not all about greedily getting gains after all.) - you cant train/inform AI on data that it doesnt have access to, thus you can't have it replace what represented that data. I think a bit more of pragmatic examples would be quant pod strategies, or highly respected michelin star restaurant secret recipes passed down from generation to generation of a single family. ok tahts all bye.