[ajafari1]

Any bets on when the majority of software development is automated? My estimate based on the exponential improvement rate of AI on various tests like MMLU and MATH is within 3 years. I know that might sound crazy, but a year ago it would have been crazy to think that graphics design jobs could be automated by an AI in 2022.

Feel free to disagree and discuss why.

[rendang]

Pick a baseline, like 1970, and call the level of coding productivity based on its contemporary languages, processes & tooling "0% Automated". Wouldn't we already in 2022 be well over "50% automated" perhaps even 80% or 90%? So we may indeed see some improvement in 3 years, but it would be a more accurate perspective to call that a move from 90% to 95% automation or something, not 0 to 51%.

Think about agriculture, for example - based on a preindustrial baseline we are at least 98% automated now. Likely also quite high for textiles and other industrial production. It often seems that in discussions of near-future automation, the vast gains in productivity seen in the past 2 centuries are ignored.

[DoingIsLearning]

> Any bets on when the majority of software development is automated? My estimate based on the exponential improvement rate of AI on various tests like MMLU and MATH is within 3 years.

An AI solution will always target bang for buck first. So _potentially_ web development or CRUD apps have the greatest probability to be automated.

Nevertheless there is a whole zoo under 'software development'. Bio informatics application, Fast trading software, medical ventilators, hardware debugging with software interfacing with real-world sensors, etc., etc.

Even if your scenario comes true there will be so much 'domain-specific' knowledge that you would still need some sort of 'Business/requirement analyst + Architect' to be able to steer AI agents into any meaningful product.

Maybe the future of Software is not in the writing but in the Designing for the specific domain. Which come to think of it is not that bad of a job and probably not that far off from what a Principal/Staff Engineer is doing nowadays.

[NavinF]

You must be using an unusual definition of automated. I’d be willing to bet $10 against your 3 year estimate if you can define it more precisely. I doubt it will happen before language models achieve human level perplexity which should happen around 2038.

Also can graphics design jobs really be automated by an AI in 2022? Cause I’d sure love to stop begging people to make icons for me lol

[riku_iki]

I think the biggest issue is that those language models are still very limited by transformer window (2k tokens, each word usually consists of 3 tokens), and there is no visible improvement over this.

Your problem, code base don't fit into 700 words, you have no luck.

[bilsbie]

Well what if it gets 99.9% automated but instead it means each programmer can now program like a team of 50 programmers, devops, pms, and qa people from today.

[ajafari1]

That's the common counterpoint, but is there 50x the demand for developers? I work in tech and I know how hard pressed people are for developer talent but it is not more than a couple fold off. I doubt companies and startups would have enough work to hire >=5x the developers. Thus, AI can be very disruptive to that job function.

[stuckinhell]

A huge amount of Software development is CRUD and that will be automated very quickly and soon.