I have no idea about whether the future of AI lies in LLMs, although it would appear not. Scaling laws should hold for the next three years or so, or should open up ways for us to do more - with or without LLMs. See this interview, for example: https://www.dwarkesh.com/p/richard-sutton
Hi Ashish, I’ve been a fan since my first class with you at SIMC and I’ve been reading your newsletters, well, at least most of them for quite some time now. Thank you for these wonderfully written snippets of your thoughts.
I had a question I’d like your opinion on; what about Ai-resilient jobs? AI is inevitable. We’ll need to adapt, but how is the job market going to be for us humans, factoring in the exponential rise of AI over the next few years? Are we doomed or are there any possible job profiles that we can look into?
A friend of mine once said construction was the answer; I disagreed. I know for a fact that we have buildings that were 3D-printed. So if even a traditional, supposedly ‘AI-proof’ job is in jeopardy, what about the rest of us? I’d love to hear your opinion as I’m looking to see how I can upskill with AI so that I’m not redundant in a few years. :)
My first step would be to ask you to define "a few years", and I mean that quite seriously! :)
Over the next couple of years, expertise, taste and empathy will end up being valuable skills, particularly in education, health and in-person networking. More than that is difficult for me to say, both in terms of other professions and over longer time horizons.
I'm aware of how broad and vague this answer is, but that is honestly the best I can do.
Here's how I would apply this to teaching (my own area of expertise): I'd consider using my experience (and therefore my empathy) to figure out what different pain-points are likely to occur in each person's learning journey, and use what little I know of AI to build better, more customized learning plans for them. My taste in knowing good learning content when I see it can be used to help make AI generated content be both better and more applicable for an individual student. I hope that this will help me stay relevant in my own field (teaching).
If you have a different answer or a disagreement, I would love to hear it!
That’s more than fair. I’m looking at the next few years (say 6-8 years) similar to the ones leading up to the mass use of internet about two decades ago. I feel like AI is a significant change that is going to split our history into a ‘before’ and ‘after’ and just like computers, it’s here to stay. The people who chose to not join in will be left behind.
I’m trying to make sure I’m not one of them and coming from a non-IT industry, I’m still confused on what th right path is. Hopefully we figure it out before it’s too late.
I assume that here you are assuming that scaling loss or the bitter truth continues to work and LLMs as AI don't plateau.
I have no idea about whether the future of AI lies in LLMs, although it would appear not. Scaling laws should hold for the next three years or so, or should open up ways for us to do more - with or without LLMs. See this interview, for example: https://www.dwarkesh.com/p/richard-sutton
I had already started watching the podcast before, just haven't completed it yet.
Hi Ashish, I’ve been a fan since my first class with you at SIMC and I’ve been reading your newsletters, well, at least most of them for quite some time now. Thank you for these wonderfully written snippets of your thoughts.
I had a question I’d like your opinion on; what about Ai-resilient jobs? AI is inevitable. We’ll need to adapt, but how is the job market going to be for us humans, factoring in the exponential rise of AI over the next few years? Are we doomed or are there any possible job profiles that we can look into?
A friend of mine once said construction was the answer; I disagreed. I know for a fact that we have buildings that were 3D-printed. So if even a traditional, supposedly ‘AI-proof’ job is in jeopardy, what about the rest of us? I’d love to hear your opinion as I’m looking to see how I can upskill with AI so that I’m not redundant in a few years. :)
Hi Febin,
My first step would be to ask you to define "a few years", and I mean that quite seriously! :)
Over the next couple of years, expertise, taste and empathy will end up being valuable skills, particularly in education, health and in-person networking. More than that is difficult for me to say, both in terms of other professions and over longer time horizons.
I'm aware of how broad and vague this answer is, but that is honestly the best I can do.
Here's how I would apply this to teaching (my own area of expertise): I'd consider using my experience (and therefore my empathy) to figure out what different pain-points are likely to occur in each person's learning journey, and use what little I know of AI to build better, more customized learning plans for them. My taste in knowing good learning content when I see it can be used to help make AI generated content be both better and more applicable for an individual student. I hope that this will help me stay relevant in my own field (teaching).
If you have a different answer or a disagreement, I would love to hear it!
That’s more than fair. I’m looking at the next few years (say 6-8 years) similar to the ones leading up to the mass use of internet about two decades ago. I feel like AI is a significant change that is going to split our history into a ‘before’ and ‘after’ and just like computers, it’s here to stay. The people who chose to not join in will be left behind.
I’m trying to make sure I’m not one of them and coming from a non-IT industry, I’m still confused on what th right path is. Hopefully we figure it out before it’s too late.
Thank you for your insight.