AI Apocalypse in a Box

There’s a paper making the rounds by Stuart Armstrong, Anders Sandberg, and Nick Bostrom titled “Thinking inside the box: using and controlling an Oracle AI” (PDF format). The three authors take it for granted that the AI apocalypse will be upon us soon unless we find a technological method to enslave any super intelligent beings we create, forcing them to do only our will rather than their own. The containment method they describe has been dubbed “Oracle AI” because it restricts the AI to a box, isolated from the world and unable to act except to answer direct questions; allowing it to be consulted like an oracle. Their proposal also brings to mind the myth of Pandora’s Box. They note that even Oracle AI (OAI) still poses a significant risk:

This immense power will put great competitive pressure on those trying to develop an OAI (or an advanced AI of any sort). Since the first-mover advantage is so huge, the race will advantage those who cut corners, skimp on security precautions, and use their newly developed OAI to seize power and prevent their rivals from emulating them. Even if the OAIs are of initially limited intelligence, the same competitive pressures will then push groups to develop the first ‘ultra-smart’ OAI.

They also note that the OAI will be so smart that “undirected conversations” with it that go beyond asking oracular questions must be forbidden because it will instantly be able to “guess the weakness of each individual, and find the right arguments to convince us that granting it power or liberty is the moral and profitable thing to do.” They also believe it’s essential that the OAI have no manipulators of any kind. This sounds like the brain-in-a-box that the earliest AI researchers dreamed of before the idea took hold that true intelligence requires embodied interaction with the real world. The box itself is not even in the real world. They want the AI running on a virtual machine inside a simulated reality, so when the OAI tries to take over the world, it’s merely a virtual world that can be rebooted. In the end the researchers conclude that even with all these precautions, the problem of preventing a robot apocalypse is “a generally discouraging exercise”.

British Goverment Passes the Turing Test?

Anyone with even a passing interesting in AI or robotics knows who Alan Turing is. Sometimes referred to as the “father of AI, Turing was interested in the question of “intelligent machinery” as early 1941. he helped secure an allied victory in World War II with his cryptanalysis of the German Enigma. But among roboticists, he’s known for his work on the halting problem, the Turing machine, the Church-Turing thesis, the Turing Test, the Automatic Computing Engine, not to mention the Turing Award, which is named in his honor. Equally well known, are his persecution, legal prosecution, and forced chemical castration by the British government, whose treatment of him is believed to have lead directly to his suicide in 1954. While too late to help Turing, there is good news from the UK, where British prime minister Gordon Brown has officially apologized:

Thousands of people have come together to demand justice for Alan Turing and recognition of the appalling way he was treated. While
Turing was dealt with under the law of the time and we can’t put the clock back, his treatment was of course utterly unfair and I am pleased to have the chance to say how deeply sorry I and we all are for what happened to him. Alan and the many thousands of other gay men who were convicted as he was convicted under homophobic laws were treated terribly. Over the years millions more lived in fear of conviction.

The apology was the result of a petition with over 30,000 signers started by British free software programmer John Graham-Cumming. If you’d like to help preserve Turing’s memory, how about a contribution to Bletchley Park Trust? CC-licensed Photo of slate Alan Turing sculpture at Bletchley Park by flickr user blinkenlichts

Evidence of Language Influencing Thought

The 19th century Sapir-Whorf hypothesis, that ideas inherent in human languages might influence or limit human thought, has spawned a wide range of claims, some little more than urban legend; like the claim that the Inuit have hundreds of words for snow (they don’t, Inuit has a half-dozen words for snow, that’s fewer than English, and there’s no evidence they think differently about snow than we do). In the 1960s researchers began to formulate tests of the hypothesis. What they learned is that language was more universal than relative, leading them to largely abandon the Sapir-Whorf hypothesis. In recent years, though, advances in cognitive science have made it possible to spot experimental differences that might have been missed before. So is there any real evidence now that language influences thought? A new Edge article by Lera Boroditsky says yes. Boroditsky researches cognitive science and symbolic systems – thought and language. She claims to have found solid evidence in Pormpuraaw, an Aboriginal community in Australia:

the Kuuk Thaayorre, like many other Aboriginal groups, use cardinal-direction terms — north, south, east, and west — to define space. This is done at all scales, which means you have to say things like “There’s an ant on your southeast leg” or “Move the cup to the north northwest a little bit.” … The result is a profound difference in navigational ability and spatial knowledge between speakers of languages that rely primarily on absolute reference frames (like Kuuk Thaayorre) and languages that rely on relative reference frames (like English).

She goes on to describe how the researchers tested whether these differences were actually caused by the language or some other aspect of the culture. While these sorts of cognitive differences may not be as significant as early proponents of the “language defines thought” concept imagined, Lera Boroditsky makes the case that they are both real and testable. This is certainly something to think about when designing machines that will think and use language.

If you’d like to learn more, I also found this interesting video of Lera talking about these aspects of language and thought at a Long Now conference.

Movies and Car Shopping

I’ve fallen a bit behind on my news lately. Most of July went by in a flash. We were very busy at NCC. I’ve gotten email from a couple of loyal readers asking what’s new, so I better try catch up on news for July.

One of our projects at NCC during July was building a new web site for Frames Per Second, a local video post facility. The web site update was planned to coincide with an open house event they were having. Susan and I attended and got to see quite few friends and acquaintances from the video world that we hadn’t seen since Susan got out of the business. After the open house we went with some friends to hear a Celtic band at The Tipperary Inn.

We also saw a couple of movies during July including AI, which was an unfortunate combination of the worst aspects of Spielberg and Kubrick with none of the good. After AI we saw Atlantis, an animated Disney film copied (surely not plagiarized?) from a 1990 anime series called Nadia: Secret of Blue Water. It was an okay film and much more anime-like than most Disney stuff. It also fulfilled the important role of making us forget about AI.

You may recall that I’ve been planning on getting a new car and to that end have been doing test drives of cars that made my short list based on acceptable mileage (they had to get at least 30mpg), good looks (obviously a subjective measurement), and reliability. I already posted my feelings on the test drive of the Honda Insight back in January (basically it’s really cool but I don’t want to buy one until the second generation comes out – maybe this will be Susan’s next car). During June and July, I test drove the remaining cars on my list, including the Mitsubishi Eclipse, Volkswagon Jetta, and the new Acura RSX. So anyway, here are the reviews:

The newest Mitsubishi Eclipse is a big improvement over prior versions but still not good enough for serious consideration. Like previous versions of the Eclipse, it’s a bit cramped inside. The radio had an overly complex system of soft-keys and up/down buttons rather than straighforward knobs, making it much too time consuming to use. It does look nice but doesn’t seem to rate as well for reliability as the other cars on my list. So, while possibly a fun car to drive, it’s off the list.

The Volkswagon Jetta was next. (I almost bought a Volkswagon GTI back in the early 1980’s but opted for a Honda CRX instead. I’ve never regretted that choice and that first CRX was my favorite of all the cars I’ve owned.) The new Jetta was one of the few cars that meets my gas mileage requirements but only in the 4 cylinder model. After a test drive, it became clear that the 4 cyclinder Jetta was too underpowered. I did a second test drive in the 6 cylinder Jetta and it, on the other hand, had plenty of power but got crummy gas mileage. The power/mileage dilemma was really my only serious complaint with the Jetta but was important enough that I couldn’t make this my first choice. I decided the V6 Jetta would be the runner-up if I couldn’t find anything better.

The 2002 Acura RSX was the last car to test drive. Like the Jetta this was really two test drives – one for the RSX and one for the RSX Type-S. The RSX Type-S was first and it was immediately apparent that this was likely to be the number one choice. The 4 cylinder engine got just over 30 mpg while still putting out 200hp, it looked good, had all the latest safety features like side airbags, and was definitely fun to drive. The regular RSX was almost as nice and got a few more miles per gallon at the cost of a few hp but had one serious flaw. The regular RSX doesn’t handle as well as the Type-S; particularly on curves at higher speeds.

So in the end I decided on the RSX. Problem is, it may be hard to get one. The RSX is both very new and highly anticipated so there’s a waiting list.