AI Versus Human Reasoning Capacity In Space Combat

[JoeSchmuckatelli]

52 minutes ago, wumpus said:

Sounds like one of the big current difficulties in machine learning.  You put in your training data, look at the output data and reinforce any positive outputs.  So as long as you get the right output for that data, the system keeps reinforcing that scheme.  For modern values of "AI", this is exactly what you should expect.

I was thinking about this recently when I got into simultaneous conversations with my son and daughter about data privacy and the algorithms that determine what choices you are presented with in music and television shows. 

My major criticism is that they actually suck.  Rather than being able to predict things that you might like or be interested in - the algorithm actually serves to only limit the choices you are shown.  The machine has no ability to think outside of the box or make intuitive leaps. 

 

I'm not all that worried about the future 

 

If you think about it - we've seen what can happen when people are either intensely stupid or intentionally stupid and decide to abandon themselves to a single source of information (curated "infotainment" and outrage on FB, for example ).  Similarly we are seeing blithely unhardened systems falling prey to ransomware. 

 

The forces that want to profit off, use or destroy the unwary are busy.  But just as 'that what does not kill you makes you stronger' was true when we were cave men... We are learning and adapting to our current environment.  As we will in the decades and centuries to come 

[kerbiloid]

Just insufficient amount of learning data. Like a chart with just three points on it.

AI should be being learned on big data, and predict not a personal, but mass things.

Too make personal predictions, it should have more data about the particular personalities, and treat them not as something individual, but as individual deviations from mass averages.

[wumpus]

7 hours ago, JoeSchmuckatelli said:

I was thinking about this recently when I got into simultaneous conversations with my son and daughter about data privacy and the algorithms that determine what choices you are presented with in music and television shows. 

My major criticism is that they actually suck.  Rather than being able to predict things that you might like or be interested in - the algorithm actually serves to only limit the choices you are shown.  The machine has no ability to think outside of the box or make intuitive leaps. 

They don't care about what you would like or be interested in.  They only care about getting you to watch/pay for something, ideally which ever makes the most profit for them.  Also it would be very hard to measure how much you "liked" previous titles (unless you regularly rated shows/reviewed them), while it is fairly easy to determine the expected amount of profit from selling you something.  I'm sure this is especially lucrative for kids shows, when you not only sell them the show, all the ads spammed on the show, and the inevitable toys/clothes/pillow cases that they will try to sell you based on that IP.

I was talking about more like things here : https://www.unite.ai/the-struggle-to-stop-ai-from-cheating-on-tests/

[kerbiloid]

3 hours ago, wumpus said:

They don't care about what you would like or be interested in.  They only care about getting you to watch/pay for something, ideally which ever makes the most profit for them. 

+1
Once I watched a movie from Kazakhstan, every next time when I watched youtube , it was full of Kazakhstan movies for a year or so. Like if I asked.

The same with agricultural equipment and so on. Like if I asked.

This just proves that google neural network has ganglia instead of brains.

(They probably mark the rare topics as "special interests" and increase their weight 100 times.)