Responses to "So, IBM Has an AI (Artificial General Intelligence)" -- Part I

The previous report on IBM's de facto AI, Watson, generated a few direct and indirect responses worth mentioning. First, The New York Times followed up the article linked in my last post with an analysis which actually acknowledged Watson is a more or less a basic AI. Whether or not this admission was a response to this blog's comments, it's interesting to see a major news organization noting this fact. The Times writer also discussed the race between artificial intelligence and intelligence augmentation (AI versus IA).

While seeing these issues discussed in the public square was undoubtedly interesting, I'm afraid the quick overview provided may have an in fact understated what is going right now, in terms of these two complementary technological paths, and also the other resource and competitive pressures being brought to bear not just on the computer industry, but on the entire human race.

I will go into these subjects more thoroughly in this blog in the near future, but for now, let me clarify.

As the Times notes, AI in Watson's league is a step towards eliminating a lot of expert consultation, at least when it comes to asking questions with relatively straightforward answers. But the greatest insights usually involve far more than just a rote answer. Knowledge certainly plays a role, and a great deal of that information can and has been assimilated into books or databases. A gifted professional may well be bringing together sensory information, "gut instinct" and a wealth of information that comes from having lived in the real world and having a deep understanding of it. Some symptoms picked up by an attentive doctor, for example, may be easily checked by a machine -- blood pressure, pupil dilation, heart rate. Others, such as subtle psychological cues, or insights made possible by a long familiarity with someone's personality and general lifestyle, may prove much harder.

Truly major discoveries, on the other hand, could be much more difficult for an unassisted, simple AI to accomplish. Something like Einstein's theories of relativity were not merely a shift in the paradigm of physics -- in order to find the answers you had to understand that the questions even existed in the first place.

You might read the above qualifiers and think that I am now minimizing the impact of a Watson. Actually, no. The ability to give meaningful, even expert-level answers in a second or two when asked a question, especially when sifting vast databases, is potentially an immense change in itself.

Consider. If Watson is that capable of understanding murky questions and responding to them accurately, then a scientist or inventor could ask for information on a whole host of questions and then receive accurate and almost immediate replies. In effect, a huge part of the trivial, mentally draining, unrewarding and unprofitable work that an elite research team has to do... goes away. Or is rather handled speedily and invisibly by the machine in question.

But this is only the beginning. As I noted previously, we already have two computers out there doing research -- sifting articles for secondary drug uses and analogues in one case, and determining the effect of each gene in a simple animal's genome in the other. Meanwhile, there are other powerful means of coming up with scientific discoveries or technological innovations, such as evolutionary algorithms.

Because Watson can understand and act upon generalized questions, it's likely that soon an AI will be able to understand and act upon other orders, such as a command to begin a new line of basic research, to focus its drug searches on specific diseases or enhancements, or to find the ideal design (evolved from its algorithms) for a specific piece of technology. It's not just whether you can dispense with hiring a new expert for your team; it's that you may have gotten a specific, seemingly major task done in a matter of hours, minutes, or seconds because your automated systems were able to understand exactly what you were asking for.

The expertise you can tap in simply searching for existing answers can be equally formidable for decision makers. A country facing shortages in its food supplies could, in a matter of moments, draw up information on foragable plants, various ways to produce more food (especially fast, high-yield and/or cheap methods), and what organizations might be willing to supply significant food (for free, for an acceptable cost, or for barter (wheat for oil, or what-have-you)). These searches might miss some options (the roots of those daylily plantings lining your highways, even in the dead of winter, for example), but at least leaders would no longer be at the mercy of the information and prejudices of the experts they have on hand. And who, really, has a host of top-notch professionals on hand in every field, for every question?

This kind of breadth and speed could critically improve decision making. But, once again, it's not only the answers you get, but what questions you ask in the first place. For instance, many researchers in the fields of human enhancement and human augmentation -- the study of how to help people be smarter, healthier and otherwise "better than normal" -- seem oblivious to related, complementary developments in sub-fields other than their own and perhaps one or two others. Even though, ironically, some of that work is going on in extremely well-established disciplines. So an AI researcher might keep up with cybernetics, especially work on human-computer interfaces, and yet be unaware of much more mature fields such as nootropic drugs and nutrients, biofeedback, cranial electro-stimulation, self-hypnosis, accelerated learning and sensory-deprivation tanks... or even the full benefits of better nutrition and cardiovascular exercise or the damage caused by sleep deprivation and stress. Or quite a few other interesting discoveries.

Lacking this knowledge can lead to some odd missteps. When last polled, for example, 20% of American scientists admitted to using a drug to improve their thinking... and remarkably, the two main pharmaceuticals employed were Ritalin and Adderall -- two substances with limited uses and well-established and often dangerous side effects. With relatively safe nootropics like Piracetam and general alertness-enhancers such as Modafinil available, seeing that many of America's scientists making such a questionable choice is surprising.

Then again, the Times' followup article on Watson in describing intelligence augmentation speaks exclusively of useful software that can assist elite scientists and engineers, rather than the more formidable option of directly improving the intelligence, learning ability and creativity of the researchers themselves. In fairness, the full extent of intelligence-augmentation experimentation may have been beyond the parameters of the piece, but more frequently this oversight results from sources who are themselves unaware of dramatic progress. Sources who may also harbor prejudices against specific lines of research, such as a dry nanotech or AI triumphalist who feels an unspoken contempt for biological or psychological augmentations.

Once again, it's a matter of the questions you ask, and how frequently you ask them, not just the quality of the information available to you. Our greatest discoveries are often made at the borders of our ignorance, not just at the pinnacles of our understanding.

Which brings us to an interesting twist to all of the above. There are many forces driving dramatic change in the world today. Some of these are new technologies, new opportunities and new competitive forces, but others are grave challenges that are coming fast.

There are really too many of these factors to sum up here, so I will only touch on a few. The intense competition of the computer industry is in many ways the quest not so much to dominate a long-standing market but to create new ones. If you look at some of the major products and/or companies to have emerged in the last two decades, you see Amazon and eBay, Google, the iMac, iPad and iPhone, the Droid smartphone, the crowdsourced software "apps" for smartphones, a host of open-source software (Mozilla, Ubuntu, Python), Playstation and XBox, blogging, e-readers such as Kindle, Facebook and Twitter, and, of course, IBM's Watson. And more.

Quite a few of these innovations were sneered at, yet computer games' revenues now exceed those of the U.S. film industry, and Facebook and Twitter have been used by enterprising, educated young people as the organizational means to overthrow two Middle Eastern governments. Further, the ability to cooperate and compete over the Internet and throughout global markets, and to exchange software based "goods" in seconds, has sharpened these competitive pressures. To return to the above list: How many of those innovations were the work of companies that were either viewed to be fading or on life support, or which had only just come into existence?

But that furious commercial battle is only one tiny part of the larger picture. The debate about whether to pursue intelligence augmentation versus artificial intelligence has for a long time missed the point... Right now, we already have intelligence augmentation, and brilliant human minds that can use it. Our computational breakthroughs, whether AI-related or not, have thus far been most spectacular at advancing research into enhancing humans -- whether by decoding the human genome, assaying new nootropic drugs, scanning the mind with improved MRIs and algorithms, putting the world's scientific journals online in searchable formats, and so on. Granted, it helps that just about any medical research is "dual-use"; as virtually any medical advance can be applied to enhancing some aspect of the human condition. Alzheimer's research equals memory enhancement, intelligence enhancement, and nootropic and longevity research. Parkinson's research equals intelligence and nootropic and longevity research. Artificial limbs mean cybernetic advancements. Repairing brain damage means advances relating to cybernetics, intelligence augmentation and artificial intelligence. And so on and so forth. And, of course, much of this work is not only a theoretical augmentation. Merely keeping existing, brilliant minds functioning at their best for a few more years effectively augments global scientific and technical research. Now imagine how much more could be done to assist those minds directly.

In other words, the scientific and technical competition existing in any number of "hot" fields, and quite a few complacent ones, could be intensified simply by augmenting the intelligence and creativity of their leading researchers. Clearly if, as of several years ago, one fifth of American scientists were taking some kind of drug to amplify their intellect, then this transformation is already underway.

The other half of this changing competitive picture are all of the new people competing -- both a broader slice of the public in countries with established tech industries, and throughout the world. Whether open-source programmers on Linux, hobbyists providing apps for iPhones, or startups emerging out of nowhere, the host of new minds involved in solving problems and/or creating new products is staggering. Now, imagine if all, or even a majority, of those minds could be radically augmented in terms of their gifts, and empowered in terms of the knowledge and resources they could tap and the ease with which they could bring products and companies into existence.

But perilous changes are also taking place in our world, which provide their own kind of "competitive challenge." The world consumes over a cubic mile of oil a year, and vast quantities of natural gas and coal. Those supplies are not only limited, but the energy required to find, extract, refine and ship them to market, particularly in the case of oil, have been steadily increasing over the last century. And our production of oil is almost certainly near, at or just past our ultimate global peak in overall production by volume (and probably well past in terms of net energy).

Rising energy prices feed through into everything, particularly in oil's case, as it happens to be a feedstock in a huge number of products, in particular almost anything made out of plastic. Rising prices and/or falling profits for virtually all goods and services puts financial pressures on everything, which is bad news in a global economic downturn as severe as this one.

Meanwhile, climate change is well underway. Some of those dire impacts supposed pessimists felt could happen in just a decade or two, such as disruptions to our food production, may already be here. Severe drought in Russia, parts of China and India and in western Australia, severe flooding in Pakistan and eastern Australia, and very hard frosts and ice storms in Mexico, southern China and some localities in the U.S. will almost certainly damage global food supplies in 2011. In countries where the average household spends 40% to 50% of their income on food, doubling food prices means economic ruin if not starvation for many, many people.

We should not be surprised that dramatically higher food prices have helped drive revolutions in the Middle East. Nothing makes people believe in change like seeing the end coming.

But ironically, being driven to the wall may prove to be our greatest evolutionary hope. When you no longer have any excuses, delusions or options, you have no choice but to change. That change may be for good or for ill, but at some point it becomes inevitable. Our mission, then, is to make the best choices we can with the information and opportunities before us, and to help provide better alternatives to others, so that when they are forced to leap headfirst into change, they choose to leap in the wisest direction.