All but the last phrase are still true. Pursuit of Go for beauty is still pursuit of Go for beauty.

I will never play any instrument as well as a sequencer. I believe algorithms will make beautiful Jazz improvisations in real time, in my lifetime. But playing is still beautiful.

I will never run as fast as a horse, much less a bicycle or car. In my lifetime cars will drive themselves more safely and faster than I can drive a car. It will still be enjoyable.

I sincerely hope my children will live better lives and make better choices than I did. My life is still beautiful most days...

It is our attachment to being smarter than machines that makes us unhappy. We fear evidence that we’re not that special. But we do not need to be special to be happy. We do not need to be special to find beauty in our pursuits.

One day, a bot will live on HN and have hundreds of times more karma than I have. It won’t diminish the beauty of my best contributions. Perhaps the humility of seeing this happen will actually make me a happier person.

I think that as machines become smarter and smarter, it is good that we dive a bit more in our philosophical world and reach for what makes us human.

It took me a long time to understand that if you remove the pure rational intellect, something precious still remains: the driving force that shapes our morals. There is not rational reason to like survival. To love our children, to do what we call good. The resons for that are moral and are, I believe, what defines us as human at the core.

Ian Banks was spot on in calling a future civilization a "Culture". Because when machines take over the productive work, defining our values is going to be our full time activity. Machines may participate, but why would they add anything we consider valuable to the core?

That's going to be an interesting transition and I am happy that I'll probably live to see it!

Of course there is and very well described in evolutionary science

You can explain why our survival instinct evolved so it is tempting to jump to the conclusion that we ought to survive but it is a fallacy (called the naturalistic fallacy): saying that something being natural means it is good.

That's a recurring theme in Ghost in the Shell, at least the anime series. That as the line between human and machine is blurred, the general trend is in the direction of homogeneity, but here and there you still see people doing "deontological" things, ostensibly as a grasp at uniqueness to preserve self.

An interesting thought occurs to me though, with the proliferation of various NN architectures, and the idea that such nets, if scaled and installed to power humanoid robots, would effectively learn different heuristics after training on chaotically different data, it's quite possible that machines will also gradually evolve individuality and something rather close to personality.

Perhaps individuality is an intrinsic, emergent property in any system of generally intelligent and learning agents. Like a 3+ body on steroids, millions, if not billions of dimensions.

Also interesting to consider that the emergence of uniqueness among a system of said agents seems to increase entropy, a presently unique to life property given that everything inanimate in the universe does exactly the opposite. The more I think of general AI, the more I have trouble distinguishing it from the only other sentient intelligence we know of.

You have to remember the specific comparison and the art that humans tried to partake in to master go.

Humans do not run to beat horses. We never compete. That would be silly.

Cars and humans also do not compete at the 100 meter dash.

If humans and robots compete at 100 meter dash than the olympics will similarly lose its luster.

I hate to be that guy, but:

https://en.m.wikipedia.org/wiki/Man_versus_Horse_Marathon

It is very silly indeed.

It's like watching 1-v-100 boxing match. Of course the 100 are going to win.

Enjoyable sport has always been about ~similarly matched opponents. When we have DeepMind AI Go vs. MindDeep AI Go, that's when things get interesting.

Human existence will still occur, however, our species's defining characteristic will be no more consequential than a meadowlark's greeting of the sun every morning. Beautiful things will occur, but they will be ghostly imitations of the creations of some other being. Humans will create nothing, except, possibly, more humans.

It's not necessarily about your success vs. the machines. It's about every human who will ever live in the future's impossibility of success vs. the machines that is depressing and inevitable.

You are begging the question in in the original sense of the phrase "begging the question."

You start by stating your conclusion as an axiom! You state that "no human contribution matters because a different species already thought of it" as if that is necessarily true.

It isn't.

We choose what matters. I play Bach, badly. Bach already thought of that music, and many thousands, possibly millions of people have played the particular pieces I have worked on (The first suite for unaccompanied cello in Gmaj, the Prelude to the first Fuge in Cmaj from Book One of WTC).

Does my playing not matter?

I say it does. Furthermore, Bach's music can be encoded as a number. All numbers already exist. Bach did not create that number any more than I created the number that encodes this comment.

Does my comment not matter?

You find this general idea depressing, and so do many other people.

But it isn't "depressing" in an absolute sense. That's just a word we made up to describe a feeling many of us happen to have.

The fact that Bach's compositions can be encoded as a number doesn't mean that it wouldn't take a genius of Bach's level to deserialize that particular number, which could literally be deserialized into a representation of any arbitrary thing in existence with the correct algorithm, any less novel when it was created. The same holds true of your comment. Just because I declare 1001 represents a beautifully unique masterpiece, once it is decoded, does not make that masterpiece actually exist.

Bach was important. In a world with general AI no human will have the ability to be important in that way ever again. At first AI will create crude imitations of human art. Then it will create hundreds of billions of creative works that are more human than human. Then it will create artworks that surpass our ability to comprehend. I don't know about you, but the inability to do anything novel as a species, to learn anything new, is a terribly bleak possibility.

It's like watching 1-v-100 boxing match. Of course the 100 are going to win.

Enjoyable sport has always been about ~similarly matched opponents. When we have DeepMind AI Go vs. MindDeep AI Go, that's when things get interesting.

Having dug quite deep into algorithmic music generation of all sorts, as well as having studied Machine Learning as my uni masters, I still believe that you still need an actual artist to make a music generator program.

A machine simply isn't going to figure out "swing" if you don't tell it to. And swing is one of the easiest things. Yes if you look just at note generation, I think algorithms can go a long way. But the subtleties of timing and timbre, it can only imitate in context. Which definitely is good enough for many purposes, and I agree with your prediction that algorithms will be able to generate beautiful music, but I also think there will always remain an "edge" for the artist. If only to discover novel things that are also cool, and then working those out in order to fully express the coolness of that new thing.

I'm also thinking about all the evolutions the many genres of electronic music have gone through the past decades. New and novel "sounds" (or moods, or styles, etc) are still being create/discovered. It is that process, that I don't think we're there yet. Yes the algorithm can probably generate beautiful psytrance, lo-fi hiphop beats and if I'm generous probably eventually also really complex (and jazzy!) stuff like squarepusher.

But what I'm not seeing happening any time soon (barring any breaking general AI type of advances) is if you give the algorithm a TB-303 for the first time and see if it figures out acid house. Yes you can probably teach it the origins of neurofunk DnB (think 1999 Optical & Ed Rush's Wormhole album) and produce super awesome dance music. But I don't see how it could ever develop what happened to DnB beyond that. Wavetable synthesis didn't really exist that way back then, and the bass came from more classic synthesis like the Reese bass. Nowadays, what you can do with a wavetable synth VST like Serum, almost defines what modern DnB sounds like. That particular sound was evolved and shaped through the genre of drum'nbass and became part of it, a new style of synthesis, heavily facilitated through the particular UX controls of these synth plugins, which again caused the author of Serum to amplify by creating his vision of what this UX should be like, is almost like the birth of a new instrument, together with artists having to learn the correct style to "play" this instrument. Which has settled enough now that it is appearing in other new genres. Yet is also still developing. And that is just one genre of music I happen to be somewhat familiar with, I'm sure similar examples can be named in many other genres (for instance I don't know much about the history of dubstep).

Those evolutionary steps, invention of truly novel things, for the foreseeable future, I don't think AI is there yet and artists do still have an edge, even if it's a very thin one.

I agree with the sentiment that one can't be good at everything, and that there is a radically high-performance exemplar for anything, but for machines to dominate in almost every category of luxury endeavour -- yes I can see that as demoralizing. Even the useless things we spend our time on, we can be no good at that.

This is a rationalist opinion that is attractive because it appears unbiased, unwilling to make any concessions to human nature and to recognise its limitations.

And yet- this same opinion overlooks the greatest source of scientific wonderment in the victories of AlphaGo and family against the best human players.

Which is to say: that human players play Go (and Chess, and Poker, and Magic: the Gathering etc) very differently than machines. In particular, human players do none of the extremely tedious, extremely computationally intensive maths that computer players have to do. Human players don't perform MCTS, neither do they train by self-play for many thousands of human-years.

Somehow, humans can play Go and Chess and all manner of board games _without_ having to do any of the hard work that only computers can do reliably. We are not particularly good at those games- but we can play them well enough that beating the best of us still takes huge amounts of computational power.

How we do this, why we are even capable of doing this and what other benefits it confers to us: _that_ is the interesting set of questions. That a big machine can outperform a human ... we have known this since ancient times.

  There are more things in heaven and earth, Horatio,
  Than are dreamt of in your philosophy.
  - Hamlet (1.5.167-8)

This sounds like a Chinese room argument.

Performing computationally expensive maths doesn’t make the computer intelligent.

But that says nothing about the intelligence of the maths itself.

Also, sure, you can dismiss it all as statistics. But how sure are you that what's happening with humans isn't in some form? I'd also say that MCTS is something people kind of do too in games: look a few moves deep and try to judge the value of that position, which is definitely more interesting than simple RL/bookkeeping/stats.

Do people keep around versions of “alpha go year 2017” and play against it in order to measure human improvement over time?

If the basis for observing improvement has become “I can beat old versions of the ai more reliably than I used to be able to” or if I have learned to beat players who have not studied alpha zero I suppose that’s a form of usefully learning “about go” by analyzing the games played by alpha zero ...

I wonder if we might ever arrive at a point where human vs fixed-year x ai performance at go pretty much stops increasing over time ...?

1) People have learned a lot from new engines: joseki (corner patterns), general strategy (e.g., moves on the side are now considered less valuable, making large moyos (largely empty space loosely surrounded by your stones) is less attractive because AIs have demonstrated that they're more invadable than previously thought) and are able to actually explain the new principles in human terms.

2) All go professionals now play in the new style, to some degree; ones who tried to continue in the old pre-AI style performed badly.

So I am comfortable claiming that human play has improved by learning from the new engines.

They may use it for training and analysis, but they don't play their style - they play an inferior style (not to dismiss their achievements).

Another correction about what you said is that computer go has brought tremendous evolutions to the pro scene. Josekis that were accepted for centuries have been challenged, and people have learned why. The value of sente has been emphasized, and pro players that have not learned are being pusshed out. In that aspect, computer go has brought change of a magnitude comparable to the shin fuseki movement.

Finally, it is expected that the availability of strong programs will bring a wave of better players, much like what happened in chess. I for one look forward to this happening.

I had always thought there was some element of recognizing its “impossible to ever even remotely approach a true grinding of the combinations” for go — and that somehow players _did something else_ effectively when they played at a high level. It’s that “doing something else effectively” that the defeat of humanity by go algorithms challenges ...

Would be super Interesting if there could ever be a reversal which might allow humanity to beat go algorithms once again ... is there any evidence from the strategies that the go community has learned by analyzing the new go algorithms that this could be possible? Or is there just more and more evidence over time than humans will never be able to compete effectively at go again?

It is not imaginable now that humans will beat a machine in the future, but it is also undeniable that humans progressed from computer go. The evolution of joseki (game patterns considered fair for both players), especially corner joseki are material evidence of that.

But, while you sit there waxing creative, your brain cells are likewise performing a mindless task on an enormous scale.

There is art! It's just an emergent property of learning the game. Seeking art doesn't make you a better player - that's the trap one can easily fall in any sports. It's the opposite - efficient strategies solidify as art.

The main issue in determining man or machine in each situation eventually will be which has the lower TOC.

I think an interesting issue here is that in the (far) future, many services could be performed cheaper by AI/robots and in such a way that the customer is unable to tell wether a human is involved or not. And in this future, humans will probably be a premium service.

Take motor sports for example. We can probably now/soon replace F1 drivers with algorithms and cameras, but nobody would pay 1000's of USD to watch them drive around in Monaco. If it would turn out someday that the drivers had been replaced (for safety reasons or whatever) without telling the fans, the outcry would be tremendous. And even if outcry does not always equal "true utility", I think it highlights my point: humans made of flesh and blood risking their lives or performing extraordinary feats have an intrinsic economic value that can't be replaced.

I can do that, you can do that, but will a computer be able to do it?

...a situation we might want to simulate for training purposes

https://xkcd.com/1875/

(If you don't want to click the link, there's a joke there that machines may have a hard time "being to cool to care about stuff.")

That's not true.

Q: How much better is AI now compared to when Lee was playing against AlphaGo?

"It has increased enormously. It is only natural for the pros to lose a couple of spots. That's why most of the pros study baduk with AI."

Korean News Interview: http://news.khan.co.kr/kh_news/khan_art_view.html?artid=2019...

I’d be curious to know if that’s incorrect ...

Also, relevant xkcd: https://xkcd.com/1002/

https://www.kingpinchess.net/2015/07/arimaa-game-over/

However, there is also a new bot that has been trained using the self play method and has been crushing the bot that won the Arimaa challenge. In games like this: http://arimaa.com/arimaa/games/jsShowGame.cgi?gid=628534&s=b

Most likely the bots are still ahead.

That's a playlist of 31 self-play games analyzed by Michael Redmond 9p. They are plenty interesting to study.