Comic Discussion > QUESTIONABLE CONTENT

Shouldn't AI constantly outshine humans?

<< < (2/2)

BenRG:
That's the thing; from what Momo said to Emily, most mobile AIs spare processing power is used up just running their chassis and sensory interpretation subsystems. As a consequence, they generally don't have much different raw cognitive power than the average human. Non-mobile AIs with processor stacks the size of an average house, have huge amounts of spare processing power and really are super-minds.

Cornelius:
For that reason, I think it might be interesting to see what May's back story is like. If she had been running in a server somewhere, as has been theorised before, I can see a couple of strips where she can make very clear what the difference is, exactly.

Aside from that, there's also the question of how, exactly, AI learn. It's not because you can have the plans ready, and measure accurately - if in fact they have the necessary sensors and software to do so - that it means they can readily outshine a trained carpenter. How to put it together is another skill. I think we need to see first how they can actually gain skills, before we can assume they will outsine humans anytime.

As for jobs that already require working with a computer I'd say again that it depends. For one thing, if they are having to work with humans on the same system, I can imagine working with or in the tables directly might well give them the equivalent of a bad head ache, as it does auditors now, from time to time.

Finally, there's also the question of, whether they want to do so, or not. It seems to me that in the comic, there's enough signs of AI trying to act like humans  - going from Momo's preference for reading a physical book, to Roko's foot issues, or their recreational downcycling. It might just be something that got in, somewhere, and now is part of their make up. There's another possible reason though. Perfection is eery, and the first introduction of a full sized, human form chassis, stressed the fact that it was rather close to the uncanny valley already. I think that outside of the people we see in the comic, there's still quite a lot of fear of AI around - as shown by Evie's monologue - and constantly outperforming humans would rather feed that fear and hate. Personally, I rather prefer the idea that they like the imperfection, and try to be as much of a person as they can. There's also the fact that they are, in fact, their own person, and not just some programme on a some machine. Individualism might go a long way in explaining why some habits (of mind) are adopted. But that, partially, ties in with learning.

Edit: there's also the fact that connecting to other systems seems not to be without its risks.

awgiedawgie:
Humans demonstrate on a regular basis that there is a world of difference between possessing the technical knowledge for how to do something, and possessing the proficiency of skill needed to actually do it successfully. Just because an AI has ready access to all the knowledge available on the internet (and as we are well aware, not all of that is useful knowledge), doesn’t mean that they are automatically proficient at any given skill. They might have a shorter learning curve, so they might be able to become more proficient sooner, but it’s not automatically a given.


To use your own example of designing a bookcase, I can do exactly that, all in my head, and I have been doing it for 40 years, so I have the necessary proficiency to build it as well. Yes, I still need to use a tape measure. But there’s no reason to assume that an AI would not need to as well. There are too many events in the day-to-day life of a person that could throw built-in measurement sensors out of calibration. That’s why engineers keep their precision instruments in cases and treat them so carefully. So I reason there would be only very rare instances where a chassis would have such instruments built in. For the average AI chassis, they would be just an excess cost without a significant return.

Is it cold in here?:
Jeph has portrayed transhuman AIs so we know it's possible.

They seem to have little ambition. Jeph said one reason there's been so little competition for jobs is that many of them are lazy. Socializing with humans seems to be a priority.

Jeremy probably out-competed humans doing the kind of work a robot arm does.

Navigation

[0] Message Index

[*] Previous page