-4
$\begingroup$

It sounds like people boast of something being "artificial" about machine learning when actually people boast that humans implemented algorithms like e.g. Monte Carlo Search (MCST) etc. I think the term AI is only to market to uneducated audience and should be banned. Machine can only look at the raw data. Any intelligence is human. e.g. when Alpha Zero beats something at chess , it implemented MCST (created & implemented by humans)and used a huge data may be too many orders of magnitude to beat humans. Nothing "artificial" about intelligence.

It sounds unimpressive when one says Monte Carlo powered x numbers of TPUS with y amount of RAM and zillion penta bytes amount of database beat a human at a game. it sounds great when you say AI beat human.

But I do appreciate the fact that the creators did not hand code any chess rules (other than legal moves) & that's a success story for machine learning.

$\endgroup$
  • 1
    $\begingroup$ I suggest you read about the differences between "applied AI" and "general AI". Moreover, about current trends, challenges, ... (chess has been proved not be a good area to measure intelligence). Finally, it is obvious that a machine playing chess is artificial, what can be doubt is if it is "intelligent" or, better expressed, the definition and subareas of intelligence. $\endgroup$ – pasaba por aqui May 8 '18 at 16:21
  • 2
    $\begingroup$ Welcome to ai.se....no offense, but this sounds more like an opinion based rant against AI...also humans are taught many things...that does not mean they are not inherently intelligent..or maybe dogs and gorillas can be a better example. $\endgroup$ – DuttaA May 8 '18 at 16:40
  • $\begingroup$ Welcome to AI! Unfortunately, I have to close this (possibly temporarily, pending revision.) Essentially, you're making an assertion, not asking a question. My advice is take a step back, focus on your main points, then reformulate as a challenging question as opposed to an assertion. $\endgroup$ – DukeZhou May 8 '18 at 18:04
  • $\begingroup$ I'm voting to close this question as off-topic because not a question in its current form $\endgroup$ – DukeZhou May 8 '18 at 18:04
  • $\begingroup$ I've re-opened provisionally, with an edit to the title to make it a reasonable question. The body of the question is still quite "ranty", but I don't want it to look like we're suppressing a legitimate question, as there are many counters to your assertion that can be backed up with facts. You might consider replacing the "terminology" tag with "philosophy", but, overall, I'd advise another rework of the body of the question so that it doesn't feel like you're shouting (all caps, etc.) $\endgroup$ – DukeZhou May 9 '18 at 19:29
0
$\begingroup$

Evidence would suggest this assertion is wildly inaccurate.

While humans devised methods such as backpropogation, which supports unassisted machine learning, the algorithms do, indeed, learn on their own, without human assistance.

This is not my area, but as someone who is working on the other side of the fence (heuristics), I can tell you that what the Machine Learning and Neural Network researchers are doing is the opposite.

The main point is, no human taught DeepMind's algorithms to play Chess and Go, they learned by themselves and have exceeded not only the most skilled humans, but the previously strongest AI's, many of which we're "taught" by humans.

I may come back and add excerpts on Machine Learning, but I'd recommend doing a bit more research, and perusing some of the recent papers.

I don't disagree that the extension of rationality via faster processors and more memory is a critical factor in the recent successes, but I suspect it's only the smallest part of the picture.

Key Point: Current Machine Learning methods are being effectively employed for problems that do not have a brute force solution. Thus the choice of Go, which is unsolvable (immune to brute force).

If you can brute force a problem, there's no need for techniques like MCTS.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.