[jacquesm]

The number of 'niche tasks' where it has had immediate application value is enormous, to the point where it will be quite a while before all of the low hanging fruit has been plucked, it definitely wasn't a nothingburger.

I've looked at numerous start-ups over the last couple of years that managed to substantially alter the landscape they operated in on the basis of the change in the state of the art due to AI/ML.

[spaceflunky]

I would also add that there is going to be INCREDIBLE resistance to AI/ML applications. Both from menial AND skilled labor fronts.

I've been following a dental AI startup. They've pivoted a couple times and I suspect it's because the dentists who would use this: 1.) don't want to make themselves obsolete and 2.) don't want to give credence to a robot opinion in the room. (e.g. have the insurance companies fighting them because what the AI bot says goes against whatever the dentist is saying).

Another prime example is the legal industry. There is soooo much of the legal industry that can be automated. We've been saying that the legal industry will collapse for decades because of this. However, coincidentally the people who stand to be made obsolete are also the people that can legislate themselves from being obsolete.

[jacquesm]

Agreed, I see this mostly as an expert-assist thing rather than something autonomous. Otherwise you can plan your sabotage party.

[spaceflunky]

>Agreed, I see this mostly as an expert-assist thing

Thats exactly what happened to the dental AI startup I mentioned. They started off with "we make dental care more efficient" essentially undermining the dentist, to "dentists, let us help you with your practice"

[joe_the_user]

AI can "almost" do everything but doesn't seem to be able to fully do very many things - the self-driving car failure epitomizes things. Health care is another "promising" field where failure has been universal.

Maybe some of suggestion type applications will work out but they don't quite seem like a revolutionization. Plus playing go, better astronomy image filtering and tweaking whatever server algorithms.

[bcrosby95]

Is there an easy way for the lay-tech-person to get a window into this world?

It feels like the internet and mobile were a lot easier to follow. But with AI/ML, most of it seems to be stuffed in black boxes that I have no way to follow.

[jacquesm]

Possibly the various kaggle image recognition challenges would be one way:

https://www.kaggle.com/competitions

but there are plenty of others there that you might be able to relate to.