Laid off? Downsized? What did you learn?

Ronen E

Problem Solver
Staff member
Super Moderator
#71
AI has to experience the scenario many times before it can learn what works and what does not.
Only if it's isolated.
BTW, it's equally true for humans. You don't become an experienced driver overnight. Additionally, some of your competence as a driver comes initially from the experience of others (e.g. driving lessons or new-driver coaching etc.).
It's hard to accept, but eventually we won't be so different from AI.
 

Ronen E

Problem Solver
Staff member
Super Moderator
#72
it will only work if all are forced to do it, so that predictability can be restored.
I disagree. Traditional programming requires predictability, AI in principle doesn't. Maybe we're not fully there yet, but it will come and I believe it won't take more than 5-10 years. Automated systems definitely have shorter reaction times than humans, never get tired or bored and can handle large amounts of information continuously.
If the introduction of autonomous cars will require that "all cars are automated" it will never happen (wet dream of traditional stakeholders), and I think those working on it and towards it understand it very well. I don't think it's a working assumption - on the contrary, these cars development assumes human drivers and unpredictability around.
The real difficult questions about autonomous cars are not the technological but the ethical ones.
 

Ronen E

Problem Solver
Staff member
Super Moderator
#73
We have to adapt to the new environment in order to survive, try to prepare and watch for the warning signs.
The entire discussion on whether AI is good or bad for someone, let alone if it's coming or not, is missing the point. The main question is not whether it's coming, not even when it's coming, but how we prepare for it and deal with it in a good way. It's a bit like climate change.

Looking at it like a threat, using words like "survive" and "warning signs", is necessary only if you assume we're each to our own, playing a zero-sum-game. Once you zoom out to a society-view you might realise that AI (like previous technological revolutions) holds a huge potential to improve the human condition. Of course it also has the potential to hurt a lot of people on the way (like e.g. the industrial revolution), but this is not inevitable, and is more a matter of social perspective - do you let "the system" alone, to work out a new equilibrium, let people compete against each other with the fittest / luckiest "surviving" at the expense of the rest, or do you plan ahead so that everyone wins to an extent?

AI or not, we've already reached the point where technically not all humans are required to work for everyone to live very comfortably. This is going to get even more extreme in the future. The reasons it's hard to see (and easy to forget), let alone impossible to implement, are in the prevailing social, political and economical systems. Basically, humanity (or individual countries) has a choice between embracing that reality and reorganizing so that everyone is reasonably well off, or keeping the current paradigm that will inevitably lead to a well-being chasm between a tiny minority that has everything and masses (billions) that have nothing and are barely surviving. It may be enticing for some of us to believe we'll end up on top, but there's a much larger probability we won't. As the trend continues, slipping over the edge will become very abrupt and there will hardly be any chance for the individual of ever coming back.
 
Last edited:
#74
Ronen E,

I agree with most of your points, the only thing I am not so sure of is:

or do you plan ahead so that everyone wins to an extent?

I may be a bit negative on this, but I have not seen many examples in human history where a society can plan their way out of a paper bag much less plan for something much more complicated.
 

Ronen E

Problem Solver
Staff member
Super Moderator
#75
@Michael_M
Yes, it's a big challenge. But to even begin coping with it society needs to first acknowledge where it stands and decide where it wants to head.
I think that Scandinavian countries are not a bad example of the state of mind I'm talking about. Australia is half-way / half-hearted.
 
#76
Is anyone inm this tread clued up on employment law? If so can I private message you a question

Kind Regards
I know enough that 99% of cases are more complicated than trying to deal on a casual basis. If $$ is a problem, seek out free or low-cost legal clinics. Most attorneys of my acquaintance who specialize in labor law will charge a relatively modest initial fee to listen to your story and advise an estimate for resolving your issue or advise you that it is not cost effective to pursue a remedy. Your local bar association will have referrals to labor law specialists for free.
 

mattador78

Involved In Discussions
#77
I wonder how well they will adapt to surprises. I think self driving cars will have problems dealing with human drivers until the government forces 100% self driving cars. An example: A coworker has auto-braking technology that slows and brakes if the car in front of you slows. He was driving in Chicago and a common scenario occurred where a driver in the lane to his right cut between him and the car in front on its way to the lane to his left. His auto-braking aggressively braked unexpectedly almost causing an accident whereas a human driver might have simply let up on the gas knowing the car was moving through to the other side. It's not that they couldn't do it, it's that humans that program the technology are bad at anticipating all possible scenarios. AI has to experience the scenario many times before it can learn what works and what does not.
I agree on that one computers will react to the more statistically probable estimate and humans are inherently unpredictable, I saw something about how in some games like chess it is harder sometimes for a high skilled player to compete against a novice player due to the fact that they do not make logical moves due to their inexperience of the game, meaning they cannot plan their moves ahead. After a few games t reverts to type but it does confuse the more skilled player.
 

Ronen E

Problem Solver
Staff member
Super Moderator
#78
I agree on that one computers will react to the more statistically probable estimate and humans are inherently unpredictable
Machine Learning is based on neural networks, which is a simulation of how the human brain works. As AI gets better and better (i.e. developers come up with better ways to implement neural networks), the practical difference between how an AI and a human will react will shrink, eventually to a point where any difference is practically meaningless.
Eventually, if an AI will fail to predict how an "unpredictable" human will react in a given instance, the majority of humans will fail in the same instance as well.
It's a little unfair to compare an AI out-of-the-box with a very experienced human. Not all humans are equally experienced. A more fair comparison would be to compare an out-of-the-box AI with an infant human. Obviously you won't let an out-of-the-box AI drive a car, like you won't let an infant, or a toddler, or a child, or even a young teenager, drive a car. Luckily, AIs will be able to tap almost instantly into a lot of experience accumulated by their predecessors. This is something humans can't really do.
 

Ronen E

Problem Solver
Staff member
Super Moderator
#79
I think that the main characteristic of true AI is something that most people don't get - AIs are not programmed how to behave, not even in a general way. Instead, they are given the ability to learn, then they are exposed to information and experiences, and they learn and arrange their knowledge on their own. One option is to hardwire them with "morals" (what's good and what's bad) and let them figure out how to maximise good and minimise bad; another is to let them learn on their own what's good and what's bad (I know, it might be scary, but if an advanced AI is exposed to the sum of human knowledge it might come up with some acceptable universal principles; or not... LOL).
 

Top Bottom