[bochen777]

Open letter to Google and Go Community after the Master/AlphaGo switcharoo, lets ask Google for meaningful change for the true betterment and enrichment of all mankind.

We are at crossroads, at a moment in time where we face these moments of truths in the turning point of history in how companies like Google handle A.I. and the manner and ways in which they go about doing it will greatly influence, shape and affect the future events of the world. This is more than just about Go. And we need to ask Google and other innovators to take a much more transparent stand and a more open approach to all of this.

Had Google played fair from the getgo. Eventually A.I. would still have beat humans. But it would have done it in a manner that was fair and square. And had been an accurate truthful representation reflective of the real progress of A.I. at the time. Instead, Google artificially and intentionally, structured the so-called show-casing, of AlphaGo technologies in such an underhanded way using tactics to manipulate and deceive human perception into falsely and incorrectly believing that AlphaGo and Deepmind and Google tech was always better than humans at each and every level and stage of the A.I. advancement and development, and that such was always one step ahead of the curve ahead of the race, so to speak... But as we all know, this was not the case, far from it. And so the public deserved a transparent honest and trustworthy development of AI and not one in which shenanigans are pulled before our eyes for PR stunts and corporate imagery and such.

The recent mystery, mystique and confusion surrounding "master" has now been officially disclosed. This "name change" was most likely motivated by the element of plausible denial-ability. Because only when after Master succeeded in beat the top pros in a series of rapid online games did Google then and only then come clean and to claim the credit. Had the bot failed early on, very likely it would have been pulled, the disclosure and connection never officially made, and Google's AlphaGo still retaining its untarnished brand and records. The using of different names and different accounts goes directly against what Google had earlier promised in its earlier statements in which it had actually went out of its way to affirmatively declare on its own volition that in the future all AlphaGo games would be played under its real name of AlphaGo and AlphaBot. False flag operations (Master had a Korean flag) are usually the specialty reserved for intelligence agencies. So unless Google is taking a page from CIA/NSA, one should never have expected this level of trickery, and there is no justification or excuse.

There is a pattern of this sort of deception and illusion creating trickery going on at Deepmind and Google with respects to AlphaGo. Everything seems like very controlled release and kept too close to the chest, like this sort of artificial illusion of always being perfect and absolutely infallible at every step along the way. Akin to gaming human perception of reality. And not at all academically nor intellectually honest or transparent. For example, had the version of AlphaGo that defeated Fan Hui played Lee Sedol first and right off the bat, there would have been very little chance of AlphaGo securing any victories and very much more probably that Lee Sedol would have won all five games. Had the March 2016 version of AlphaGo played Ke Jie at the time instead of Lee Sedol it is again more likely that Ke Jie would have won many more games. Had AlphaGo and Lee Sedol immediately continued to play five more games after the first five, again it is likely that Lee Sedol would have obtained a higher win rate. Had AlphaGo actuallyed played Ke Jie last year, as previously agreed upon, there was a chance that Ke Jie could have won against it at the point in time. Same for Master. Had Master of right now played in six months or a year later's time any Chinese or Korean bot, Master would also have lost to it. In fact, the reason Google used the new name and kept the identity such a secret was because had Master totally failed to win in the test games, probably its true identity would never have been revealed to the real world.

To Google's credit, what it has proven is that AI Go , when done correctly, will invariably and inevitably far surpass human Go. It appears that is no longer the question and probably hasn't been a point of contention for quite some time. And truthfully, having a machine that plays Go at "God" like levels will help boost all of Go, even the raw enjoyment of human games. So regardless of intent, this push and development has done some good for the Go community and serves to open the door for others to follow...

However, the Google founders claim to be for openness and democracy and for leveling the playing field and not doing evil. If they had spent even a tiny fraction of the time and resources and effort to do whatever it could to in some way open source AlphaGo code (I had contacted Deepmind developers last year, they stated it was possible but would take some work) to be made available for free for the masses at large and not as much on this ridiculous string and incessant series of escalated PR Stunts then humanity itself would be far better off in both the short term and the long run.

I call on Google to do the right thing, to be more transparent where possible, and to in general be a better citizen of the world, more humble, not as proud, and no more of these sleight of hand trickery shenanigan. If Google would be willing to consider open sourcing whatever it could of AG to the community at large, this would be a solid first step in the right direction and on the right side of history.


=====


So now its proven... Master == AlphaGo.

The name.... many folks thought they named AlphaGo because of Alphabet (Google's parent company, which is also Deepmind's parent company) but rather the truth was it was named "Alpha" in the sense of the Alpha Male... the male that dominates all other males. Now with the name of Master, this not only continues the bragging but also hints of racists (slavery) and brings the symbolic elitism onto another new level not so different from the Nazi Germany regime. (first it was Alpha male, now it is Master race connotations)

We now know for sure that Google is not in the spirit of sportsmanship and indeed with only these five games and nothing else (no source code, no open to play for a fee, etc) it is intellectual misery surpassing the level of what IBM did with DeepBlue when computers first beat Chess players 20 years ago. One would think we have progress into a more open fair society but guess not really.

Especially the name change. Master = AG, then the name change and the mysterious fashion of how this was conducted and etc was entirely unnecessary and unbecoming of a Google that used to vow to "do no evil". (especially since they promised if they ever do play online it will be under the name of AlphaGo, so this bait and switch would make it all double deceptive and hypocritical; at the very least Google is guilty of intentional misrepresentation but although in this case probably not a crime in US/EU/UK, but certainty goes a long way towards being unethical and of bad spirit, almost akin to playing with fake accounts, or having dup accounts)

Google should offer the Go community a full unconditional public apology of the utmost sincerity and generosity.

Previously AlphaGo representations openly denied that another bot was AG, this time around they "no comment"... and since it is so much stronger than anything else out there right now, even Ke Jie stated much stronger than AG back in March, it is a reasonable inference. The fact that right now Master played 50-0, nearly all the top players except for Lee Sodel, (why would he if he already got beat by AG) really lends credence to the theory that Master is Google related and does not bode well for the future of Go due to Google's underhanded tactics and shenanigans.

I say boycott Google.

Edit: We had so many naysayers on here saying that (wrongfully predicting) there was no way Master was AlphaGo etc..... Seems like Google intentionally fooled a most part of the Go community and should be utterly ashamed of its actions, in direct contradiction to what it earlier had affirmatively promised on its own volition that it would play under "AlphaGo" and "AlphaBot" and instead it did the "Master" bait and switch...

[KaiserJeep]

In a world where all of us are (in most cases resentfully) interacting with increasing numbers of AI systems in the business, government, and entertainment arenas, I find that this "scandal" won't even give rise to a yawn.

You see, real human people are obsolete. We recently explored this relentlessly dystopian concept in this thread:

http://peakoil.com/forums/obsolete-t73001.html

I recommend that you read the thread, view the video, and think about your (obviously heartfelt) complaint in the new perspective just afforded. You see, in the greater context of this site, we are daily contemplating TEOTWAWKI, and most of us barely know that Go is a game played on a wooden board with black and white pebbles, and could not care less.

In the interest of full disclosure, I use Gmail and possess several Android mobile devices, and I have a vested interest in the continued health of the Google corporation.

[JV153]

In the interest of full disclosure, I use Gmail and possess several Android mobile devices, and I have a vested interest in the continued health of the Google corporation.

I once logged into Google g-mail and accidentally logged into somebody else's account. This occurred, not once, but twice. Also, a librarian told me, several weeks before this happened, that I had a g-mail problem.

[Plantagenet]

You see, real human people are obsolete.

Not really.

The problem of AI systems becoming more "intelligent" then humans has been considered for decades. A solution has already been proposed.

Asimov's three laws of robotics first appeared in a 1942 SciFi story. They state:

1. A robot may not injure a human being or, through inaction, allow a human being to come to harm.
2. A robot must obey the orders given it by human beings except where such orders would conflict with the First Law.
3. A robot must protect its own existence as long as such protection does not conflict with the First or Second Laws.

As long as programmers hardwire in some version of the three laws into robotic AI systems, we shouldn't have a problem.


Cheers!

[JV153]

Wait a few seconds Plantagenet, not so fast.

1) All good robots obey the law.
2) If you obey the law you are a robot - with perhaps some kind of overtones of a computer program with mind narrowing results.

[Plantagenet]

Wait a few seconds Plantagenet, not so fast.

1) All good robots obey the law.
2) If you obey the law you are a robot - with perhaps some kind of overtones of a computer program with mind narrowing results.

You seem to be confused about the difference between humans and robots.

Robots are machines. The AI that controls them can be programmed when it is created so that they will obey the three laws of robotics or any other laws that the programmer cares to program into the AI.

People are not machines and people have free will. There is no computer code controlling human decision making. Some humans choose to mostly obey the laws and some choose to mostly disobey the laws.

Cheers!


Humans and AI robots are different.

Cheers!

[JV153]

Robots are machines.

Yikes, so what ?

A lever is a machine. A femur and tibia form a lever, so you probably don't know the basics of machinery.

I find the current education system somewhat like training robots, with not-so-hilarious results, that's actually what I was getting at.

I realise that children might post on here - so .. but since you've read Asimov I assume you're an adult.

[JV153]

-- feigned cheeriness (Cheers!) See HAL's 'Bicycle Built for Two' defense.
-- repetitive non-contextual activity, ie programmed language parsing
-- inane/silly post-script headlines.

Feigned cheeriness is quite common, I'm a bit of a grump myself.. but jeez.. a robot could also
refer to a boot-licker.

[JV153]

Boot lickers read off the Masters' script, ie the program or paradigm. It's a collection of coded instructions, conditional and hierarchical action pathways.

[Plantagenet]

Robots are machines.

Yikes, so what ?

So machines---even machines with AI---are fundamentally different from humans in the way they are programmed/educated and in the way they subsequently make decisions and in they way decide to follow (or not follow) rules and laws.

Get it now?


Cheers!

[Outcast_Searcher]

To Google's credit, what it has proven is that AI Go , when done correctly, will invariably and inevitably far surpass human Go. It appears that is no longer the question and probably hasn't been a point of contention for quite some time. And truthfully, having a machine that plays Go at "God" like levels will help boost all of Go, even the raw enjoyment of human games. So regardless of intent, this push and development has done some good for the Go community and serves to open the door for others to follow...

And to me, this is what matters.

As a fairly strong chess player with an avid interest in computers and in A/I, I followed the development of computer chess fairly closely from about 1979 to past when Deep Blue defeated Kasparov.

There was endless carping from the chess world about how the computers were "cheating". About how it was "unfair" because computers used search and didn't really play chess. And endless denial that a computer would ever be an "A" player, then ever be a master, then ever be a Grandmaster, then ever be a world champion.

I got yelled at by more than one grandmaster for "daring" to disagree with him that computers couldn't possibly be world champion. I was told to "respect the ability of the grandmaster" and not to offer my opinion. I cheerfully replied that "I DO respect the grandmaster ability - in chess. However, I am a professional computer programmer that has written more than one chess program which did "new" things in the sense of playing like a person -- because that interested me. So how about respecting my ability as a computer programmer?"

The stunned look I got in each case was priceless.

Go was a more difficult problem than chess (it was fairly easy to assign numerical values to various aspects of a chess position. When programmers asked strong Go players about this, the players talked about concepts like "thickness", which as a weak Go player gave me NO CLUE about numbers), and its relatively quick solution (relative to beating humans) is impressive, at least to me. But if a computer will be driving my car in 5 to 10 years, I would expect a computer to beat me at pretty much ANY board game that programmers put a lot of effort into. So what? I can't hope to evaluate hundreds of millions or even billions of board positions a second. So be it.

.....

Machines will continue to out-compete people in many intellectual areas, just as machines have out-competed people for a long time in various physical tasks.

People (i.e. humanity) will need to adjust. I suggest thinking about adjustment instead of raging about Google being "unfair" -- it will be far more productive.

[JV153]

There was endless carping from the chess world about how the computers were "cheating".

You sound surprised about this.

Btw, taking notes during a chess game isn't allowed, unless somebody notices it.

Incidentally, the last time I was engaged in an education program, the lecturers all of a sudden dropped the requirement of not having cheat sheets, even entire books were allowed.. and then I started to get worried. (because where we were going, those things wouldn't help at all)

[JV153]

So machines---even machines with AI---are fundamentally different from humans in the way they are programmed/educated and in the way they subsequently make decisions and in they way decide to follow (or not follow) rules and laws.

Get it now?

Well, for starters human brains are electrochemical in nature, that's one major difference. I was referring to programming, particularly the TV guide, which I had a lot of as a kid.

[Outcast_Searcher]

Btw, taking notes during a chess game isn't allowed, unless somebody notices it.

Yes, that sounds familiar, from my tournament days. The idea is that you couldn't formulate plans or strategies and write them down for later reference.

But the computer isn't "taking notes", it's evaluating positions and calculating what to do from those evaluations, just as people do. The computer does it faster and does it differently than humans (humans think about things like pawn structure, piece placement, weak squares, etc. and computers count and do math based on such features, but the basic concept is the same.

The issue I have with computer "cheating" is that, IMO, computers shouldn't have access to online resources, giant opening books, or databases with literally, all solved positions for when there aren't many pieces left on the board. (7 or less pieces was completely solved, last time I checked. A quick check, and Wiki says 7 is still the current number.)

But chess computers continue to get stronger, so I suspect such cheats wouldn't even be needed anymore to beat a mere human, but that's just my opinion.

For what it's worth, I never used any of that "cheating" stuff, and had the programs look deeply at relatively few positions, since I was interested in real "artificial intelligence" -- or in some way emulating a human approach to problem solving.

[JV153]

But the computer isn't "taking notes",

Once I took a memory stick out of a computer and forgot it wasn't there.. jeez. I must have flipped when I was supposed to flop.

[bochen777]

It is the true democratization of AI that ensures it will be developed in a safe manner, something like Mr. Musk with his OpenAI is doing. While I can understand why Google would not wish to share the AlphaGo source code, (their claim is that it was too embedded into their core AGI that to separate it out would take a lot of work, too much time that they felt wasn't what they wanted to do) what is less understandable, given their stances on "doing the right thing" (official Alphabet slogan or motto) is not giving public access to a compiled version. And sure, some may say that strikes fear of reverse engineering, then they could simply charge the public for access to AlphaGo via remote server plays online and whatnot. This takes away each and every excuse. The only reason they haven't done it is because they don't want the dilution of their power or prestige or mystique or whatever one calls it. This goes completely contrary to the sort of openness and democratization of AI for the true betterment of all mankind that was promised. Sure Go is just a game, but the argument can strike both ways, it has been over a year now since the Lee Sedol games, and the "Master" event notwithstanding, Google has done nothing since then, secretly working to improve the AI, but no source code, not even of an old AG13 version, no more papers, nothing.... Given how they are keeping this so close to the chest for what amounts to a board game, imagine when it comes to real applications and when the stakes are higher what treatment they will give the public and how they will handle it all? To me, these patterns and trends don't paint a picture of optimal or ideal transparency and openness and certainty doesn't live up to the stated goals and ideals. Google being of such power, and the fact that absolute power can corrupt absolutely, its even more important than ever that they set the right kinds of examples, and in my opinion so far they have gone the opposite way.

Yes, I realize Google doesn't owe anyone anything. Legally speaking, every publicly traded company's first responsibility is their shareholders. But when they (Google) decide to present to society and the public at large this image of them not just being any other company but actually a real force for good, an entity that seeks the betterment of all humanity and mankind, and when they get the public (or in this narrow case at least the Go community at large) to "buy in" onto that idea, to go along in masse and cheerfully and willing participate in good faith to help prop up Google's initiative, and to help garner goodwill and to publicize and shine a good light onto Google and its projects and whatnot, AND then Google ultimately stood to benefited and did indeed benefited from all of this, then I think the public is equally justified in calling out Google when they don't actually live up to their end of the bargain and when they fall short of their stated ideals, their bargain with society. They cannot have their cake and eat it too. And for a company like Google, with them being a pioneer in AI and their ambitious endeavors into artificial intelligence may soon be responsible, however directly or indirectly, for the replacement or obsolescence of so many different jobs in so many different fields, potentially ushering into a new "intellectual revolution" and reshaping the structure of the workforce and indeed the human face of society itself, perhaps even redefining what it means to be human, I think it is imperative on us all that we keep a closer watch on these companies and the way the conduct themselves, to hold them responsible and keep them honest and accountable, and to publicly call them out when they are not living up to their promises that they made to society, because yes it may very impact soon have a material impact on us, our daily lives, and the world as a whole.

[KaiserJeep]

Look, I have read your tirades as you posted them, and I know you are sincere. These are the Top Ten tech corporations in the Fortune 500:

1. Apple
2. HP
3. IBM
4. Amazon
5. Microsoft
6. Google
7. Intel
8. Cisco Systems
9. Oracle
10. Qualcomm

Understand that Google and Apple together flat out own mobile computing, Intel and Qualcomm make hardware, HP and IBM are growing doing systems integration, Oracle and Cisco are losing ground on systems integration, and Amazom and Google are duelling over online retailing, with Google way behind and pretty desperate to close the gap.

That's what it's all about, money. The corporate bottom line. You are trying to attribute emotions and agendas that are not there. When you are falling further behind, you do things and say things that are all aimed at making more money. It's as simple as that.

Full disclosure: I retired in Silicon Valley in 2015 from HP, with 36 years service.