AI Panic!

Until recently, the term “artificial life” referred almost exclusively to computer programs, or maybe robots, but the progress being made by synthetic biologists building real cells out of real biological material means that the field is on the cusp of a major step forward, and that this conference has the potential to be a real breakthrough event.

Although this sounds quite impressive, it is not likely to be happening. The announcement of the ALIFE XI conference at Southampton later this year is full of optimism and non-achievement at the same time: Scientists get ‘closer’ to new forms of life, and they try to understand living systems. With the movement to biological cells the focus of the research shifts from a generalising, top-down approach to a bottom-up method: it is a very long way from a living cell to an organism that is remotely capable of what a robot or a computer is today. The state of the art in ALife research is quite far from (and not really focused on) the creation of intelligence. Long way to go still: AI Panic Level drops by -0.1%.

The US company DIAGNOS has made a deal with US Gold to use its technologies in the search for natural gold and silver reserves in Mexico. US Gold is paying $230,000 CDN for this service and will get some juicy bonuses if some precious metals are actually found. The “AI” behind the technology DIAGNOS applies, however, is quite a straight forward approach, in which maps (satellite images, topography, geology, etc.) are weighted and combined into a “prospectivity map”.

This technology might be used to dig holes, but is not ground breaking on the AI side. Statistical learning has been around in for quite a while now and there is lots of research on it. However, it is nice to see some of that research actually applied for useful activities, like exploiting natural resources. The slow speed in which actual AI research results are adopted by commercial companies (possibly with the exception of computer vision, i.e., face recognition) can be linked to the often difficult to understand algorithms and almost always impossible to predict outcomes of them. This inherent complexity of AI algorithms (especially of those that aspire world domination!) prevents fast adaptation by the industry and thus limits research to academia somewhat.The algorithms used by DIAGNOS are no exception of that. There is no danger of them (or any other similar algorithm for that matter) becoming conscious, thus no need to panic: As an example of state of the art applied AI, it shows how far we still are from doomsday events, therefore the AI Panic Level changes by -0.1%.

Welcome human friend!

This site is dedicated to research and unveil the perils, imminence and probabilities of a — possibly hostile — takeover of the world through artificial intelligence. I will stay on the lookout for you and post articles, research papers and break-throughs of everything that could affect this danger.

I will attempt to classify the threat value of each event into a number that indicates how much more or less likely it is that a hostile takeover will happen. Consequently, the event of this site appearing will also get a probability value, namely of -0.1%. Why negative, you ask? Well, as there is only little agglomerated information on the danger of intelligent machines (we are talking non-fiction here!), this site can (probably) help to drag this danger into the public eye and help sharpen it to avoid said takeover. This might sound a bit self-important, but it is not: As long as this site is unnoticed and ignored, so is the probability value: As long as no one cares, the measure is irrelevant anyway.

Of course, these values seem arbitrary at first, but I hope they will give a feeling of relative importance of events. My subjective feeling, for sure, but I will try to keep it as objective as possible and not drift into scaremongery. So, why the negative theme of hostile takeover and all then? For one, there is already good information on the “Friendly AI” side out there, for example from the Singularity Institute, of which I definitely will talk about a lot in future articles. Second, the danger of a “false negative”, or type II error as statisticians would call it, is much more critical when one assumes future AIs (and probably emerging consciousnesses) will be friendly: We have to be wary about a future HAL. If the hypothesis this site follows turns out to be false on the other hand, we don’t lose much. Okey, we shut down a super-friendly world improving computer, but that is better than being <insert gruesome science fiction outcome here>, isn’t it?

In this spirit, stay informed (you know where, now)!

PS: Although I’ll write about this in English, this is not my mother tongue, so it is very likely that horrible typos, adventurous sentence structure and twisted grammar will appear. Feel free to notify me of anything thereof. Preferably not in the comments to minimize the exposure of my failed verbiage ;-).