What's the future of jobs with the rise of Ai agents?

AI Discussions
1

The last agent launched by China made wondering about our future as Data scientists or AI engineer,If agents can do the work on computers independently without our assistance than what are we doing? Some people say that Ai will make new jobs but i don’t see that,what i see is that many jobs will be automated am I wrong? please share your point of view as an expert

2

For the moment I see no problem.

Relying on LLM is like relying on a drunk guy in a bar who has no hearing or sight but has somehow had enough time to braille-read the universe of text produced up to now.

It can be useful as a data-retrieving secretary, no doubt about that, but you still have to check what is on the desk.

“AI” in specific, controlled tasks where you can perform proper quality control, however, is different.

3

I think it’s mostly about scaling down. The number of jobs will go down, but there definitely needs to be human touch. I run a VFX company and see that certain departments in the industry are scaling down at a massive level due to AI. Where 100 worked earlier, now just 10-20 people will be required.
I think the only way to stay ahead is to keep up skilling and learn how to use AI.

5

There is an interesting article post written by a ML engineer developper who quit his job few weeks ago sharing exactly the same thoughts about AI agents and intelligence. Interesting point of view.
here is the article agilley(dot)github(dot)io/faang-blog.html

6

A drunk blind guy may still help a person avoid getting beat by an aggressive situation.

7

Can you verify the URL of the blog again, please? Even changing the (dot) for a dot, it’s a 404.

8

sorry for that, just add th letter “j” at the begining
jagilley[.]github[dot]io/faang-blog.html

1 Like
9

More on the foolishness of trying to have the current crop of LLMs write code, as they are not fit for purpose and unable to escape the constraints imposed by complexity theory in any case (the article does not yet include the idea that they are also abysmal at combinatorial problems, which should be the basics element to master really)

By comparison, efforts at “machinic code generation” have been going on for some time (it’s really a particular application of search & optimization), the same issue of CACM has this article on that particular subject: