In the 1 August 2015 edition of NewScientist (paywall), Martin Rees, the UK’s Astronomer Royal, suggests we’re more likely to encounter a species of Artificial Intelligence than a, well, biological species:
Many people now believe that machine intelligence will eventually surpass human capabilities. Even if this is centuries away on Earth, it is clear that technology advances in an instant compared with the Darwinian selection that led to us.
We should accept that the era of organic intelligence is relatively short, and will be followed by a much longer era dominated by inorganic intelligences. Humans and our intellectual achievements will be a mere precursor to the deeper cogitations of a machine-dominated culture.
Doom & gloom, and, at first glance, it just seems so … inevitable.
Abstract thinking by biological brains has underpinned all culture and science on Earth. But this activity, spanning tens of millennia at most, will be a brief prelude to the more powerful intellects of the inorganic, post-human era. Evolution on other worlds orbiting stars older than the sun could have had a head start. If so, then ET is likely to have long ago transitioned beyond the organic stage. So it won’t be human-like minds that we are most likely to encounter, but machine intelligences.
It honestly gives me a sense of ennui. But something rings false here – at least enough of a ring where we may still have a fighting chance. And it all begins with … slime molds. Remember them?
Samir Patel of Archaeology Magazine writes a report on how the Romans might have designed their transportation network:
… Physarum polycephalum, consists of a single large membrane around many cell nuclei, and has drawn the attention of a wide range of scientists because of its uncanny ability to solve almost impossibly complex computational problems.
Now, to review: Computers are logic machines, which is to say, computer scientists and mathematicians are well aware of problems which are not really susceptible to … computers. We’ve discussed this before in connection with Yanofsky’s THE OUTER LIMITS OF REASON: WHAT SCIENCE, MATHEMATICS, AND LOGIC CANNOT TELL US, or you can buy and read his book, or you can look up the P=NP problem. Assuming someone resolves the P=NP problem to mean that such problems cannot be solved, this defines a class of problems, some quite important, that a computer cannot solve.
AIs are, and most probably will remain, here and in outer space, computer-based. If a slime mold can solve problems a computer finds, at best, difficult, well, what problems are resolvable by biological intelligence of human magnitude, whilst opaque to AIs? This includes problems which might be key in a competition between biologicals and AIs.
I think Dr. Rees hasn’t really thought about the limitations of AI; after all, the potential is the exciting part to discuss. But fundamental limitations imposed by the basic hardware gets less attention – but will undoubtedly shape the potential and ambition of the AI.
That is not to say we won’t encounter cyborgs, which is a melding of biological with technological. The same issue of NewScientist includes an article (paywall) from Hugh Herr, a prosthetics pioneer:
Tell me about your bionic legs…
I have a company that produces what I’m wearing: the BiOM Ankle System. For the first time in history we’ve normalised walking speed and its energy cost. In other words, if you simply measure a user’s speed and metabolic energy expenditure, you can’t tell whether they have bionic legs or biological legs. That’s especially important because conventional technology used on people with leg amputation makes them limp, which causes musculoskeletal stresses that lead to joint disease and many other secondary conditions. True limb bionics eliminate limping and solve these very costly secondary conditions. Typically when we fit the BiOM prosthesis to a person, if they have hip pain, knee pain or back pain it is reduced in days.Could such bionics benefit people in general?
Actually, we have developed bionic technology for people with complete biological limbs. Last year, we were the first research group to build an autonomous leg exoskeleton that significantly reduces the metabolic cost of walking to a person without a leg condition. It’s an artificial calf muscle, which supplies about 80 per cent of the power to walk. So a person with a normal physiology could put on these exoskeletons and walk using substantially less energy. …This sounds incredibly futuristic…
Bionics is just getting started. What I’m wearing here is going to be laughable 20 years from now – absolutely laughable. So if you think stuff is cool now, it will become extraordinary – and disability will end, I’d say, by the end of this century. And I think that’s a very conservative statement. At the rate technology is progressing, most disability will be gone in 50 years.
But these are repairs and assists to the biological, not replacements for the intelligence – and thus not threatening to the biological.
