Hacker News new | ask | show | jobs
by snowpiercer 1408 days ago
Wow. ..I'm really surprised by this comment..what do you mean by "farther away than ever" . I just recently started learning to code and I really want to go into machine learning ..does it mean there isn't any major innovation right now in the field ?
2 comments
Traubenfuchs 1408 days ago
Modern ML/AI became a fully accepted, mature part of the "production software engineering world", it's a solid career choice, from everything that can be predicted right now. Existing, non-startup companies are using ML/AI to solve real problems and its buzzword powers are shrinking every day.

But regarding autonomous driving, there seems to be stagnation. We now got cars that can "autonomously drive in most conditions but required a driver at the wheel at all times" for many years now. Tesla FSD updates seem to be incremental at best. Is there any company left still planning 100% autonomous taxis that can get drunk people home? I don't think so. And that would be the fucking holy grail of the transportation business. There are about 18 million taxi drivers in the world and most of them would lose their job, saving customers and companies billions.

link
occz 1408 days ago
That's probably not what they meant.

What I interpreted it as was that the current iteration of machine learning will not be able to satisfactorily solve full self-driving cars.

link
Traubenfuchs 1408 days ago
> That's probably not what they meant.

It is, somewhat. The current generation of software and hardware was pushed to its limits by the smartest AI/ML engineers in the world and the best we got is Tesla FSD which is in no way ready to replace taxis and still needs someone at the wheel at all times.

100% self driving taxis for drunk night crawlers are not on the horizon anymore.

link