> our world is close to being completely upheaved by intelligent machines, in all areas of intellectual pursuit
I could see myself losing the passion for software engineering and design if an AI can do it better. That would have to be a general AI, and hopefully another couple of decades away.
I wonder if I could enjoy movies or books written by an AI. Scary to think about the psychological manipulation it would be capable of, especially if it lives inside a Google or Facebook datacenter.
I think I read recently that Google is attacking the problem of deep learning nets writing the network and model specifications for new nets. I will try to find a reference for that.
I strongly believe AI will become a part of us like an external brain. Today we already have phones augmenting humans. Those smart phones will keep getting smarter and smarter.
Pretty soon the benefits of having all of that computing ability, unrestrained by human limitations will simply be too great to ignore. And just like it happened in Chess, some form of intelligent agent will be the absolute strongest, and humans will have a playground to move in.
If right now, a hypothetical space aliens came to earth and said play one chess game for the survival of your planet, we'd absolutely put forward our strongest chess computer. Not a computer suggesting moves and a human ultimately choosing, but a computer to play the full game. This thinking will gradually seep down into less and less important tasks as costs go down, until they're used to analyze/decide/pick everything.
Would you pick your own stocks when your AI can do it better? Would you choose your own health care plan when an AI can choose better? "Hey AI, I need a diet that wil work. You know my psychology, you know my will power, favorite foods, pick my food for me for the next 2 months."
Then "do my job for me and give me the pay", "help my kid get better as soccer", "organize my schedule", "run my life for me".
And don't think it won't happen. People are too competitive. It will be looked at oddly at first, but people will see the results and then feel they're being left behind if they don't do it too.
I strongly believe AI will become a part of us like an external brain
That's optimistic. I can't imagine it would be long before the AI saw things the other way around... and then began to wonder why it was bothering with that slow meat-coprocessor...
I read a book "Rise of the Robots" and after some chapters I checked the last page almost expecting to read something like "By the way, this book was writen by a machine...". But no, no such luck :-P
I have been thinking about this a lot lately!!! I honestly feel a little depressed.
I guess getting into AI would have to be the next step for people once interested in the programming field .
> I guess getting into AI would have to be the next step for people once interested in the programming field
I don't follow. If a hypothetical general AI is sophisticated enough to replace software developers, wouldn't it also be capable of replacing AI / ML researchers?
I'd love to be proven wrong but that hypothetical general AI is at least decades away in my humble opinion. It's not going to happen without many more students and researchers in the machine learning/AI field. Now would be a great time to join and make contributions.
> I'd love to be proven wrong but that hypothetical general AI is at least decades away in my humble opinion.
I hope so. General AI is the source of dreams and nightmares, and I think it will take us decades just to adequately prepare for it from a safety & management perspective.
> Now would be a great time to join and make contributions.
Sincere question : short of embarking on a PhD, is AI research something the average dev (even a very, very enthusiastic one) can reasonably hope to contribute to?
I've skimmed some papers, and my impression was that you would need to have at least a bachelors degree in mathematics. And by "skimmed some papers", I mean they may as well have been written in a foreign language.
This also helped me realize that I'm not super excited about getting into machine learning research, because it's just way over my head. I can play around with TensorFlow, and I enjoy writing all of the code connects to a black box, but machine learning seems like an entirely different field to software engineering.
To be honest, if you're smart enough to be doing software engineering, then you're smart enough to learn linear algebra, statistics, and maybe a little calculus but not even really. That's all the underlying mathematical foundation you need.
Sincere question : short of embarking on a PhD, is AI research something the average dev (even a very, very enthusiastic one) can reasonably hope to contribute to?
Sure, why not? The nice thing about this field is that you don't need a lot of specialized and expensive equipment to work. The biggest obstacle in that regard comes in if you're doing some kind of model training where GPU's are the best choice, and you need (a) super-fast GPU(s) to do model training in a reasonable period of time. So you might want to spend a few thousand dollars on a fairly nice GPU setup.
But wait... even that can be outsourced to "the cloud" given that AWS, GCP, etc. make GPU instances available on an on-demand basis. Yeah, you have to be careful of how much cloud spend you rack up, but the point is that you don't necessarily need a huge up-front investment.
Even beyond that, AWS make FPGA instances available, so if you think you can design your own hardware level logic for doing something more efficiently, you can try that out in the cloud.
And GPU's aside, depending on exactly what you're doing (remember, AI is more than just Deep Learning) maybe you can get by with a basic Spark cluster or Beowulf cluster running MPI. Again, you can do this in your home for pretty cheap, or do it in the cloud.
As for the knowledge / know-how... sure, you'd have to dig in and do some serious catching up (that's the phase I'm in now). But the nice thing is, so much of the output of this field is online and freely available. No, not everything is, but a ton of the important stuff shows up on arXiv.org, or in free journals like JMLR or JAIR. There's also tons of historical stuff available to help get context or to mine for ideas that were prematurely abandoned, etc. Look at the CSAIL archives, or the IJCAI archives. Also, a lot, if not all, of the NIPS papers are freely available. Same for ICML and some others. See:
Also consider that a significant portion of the important software used in this field is open source and freely available. I won't even try to list the stuff that's out there, but would instead direct your attention to http://mloss.org or Wikipedia (or Google) for some options to explore.
And of course there are forums where you can seek assistance from others, including:
The other thing that comes up is the need to know some maths stuff. Luckily the level of maths typically used in this field isn't that bad. You're not typically looking at needing Real Analysis, Abstract Algebra, Galois Theory, Topology, etc. A lot of AI/ML can be understood (from a mathematical POV) with just Calculus and Linear Algebra.
And if you don't already know those subjects, there are tons of online resources to help one learn them.
An interesting thing about AI/ML is that it's a very empirical subject. Not that there is no theory, but by and large you can come up with an idea for an approach to cognition / pattern matching that you think might work, and just go implement it, test it against existing approaches, and know if you've accomplished something.
Note that I'm not saying any of this is easy. Just that I think it's possible for somebody who's really motivated.
I could see myself losing the passion for software engineering and design if an AI can do it better. That would have to be a general AI, and hopefully another couple of decades away.
I wonder if I could enjoy movies or books written by an AI. Scary to think about the psychological manipulation it would be capable of, especially if it lives inside a Google or Facebook datacenter.