💡 Visions of the Future

I don’t remember exactly when, but should be around 2006-2009, I was thinking what to do next in life and science. I never treated my PhD studies too seriously: I had a strong feeling that top-class research simply can’t be done within the process designed to provide you with a doctoral degree and place you onto a certain career track. You have to do a decent work, but your end goal and strict deadlines force you to minimize risk, while the "real science" is all about risk and ambition.

Back then, it was obvious that the most exciting fields are going to be computational biology and AI. Now, I am reconstructing my feelings from memory, and such things can be way off, but I remember them approximately as follows. Computational biology looked cool but my knowledge of biology was of a high school level at best, and I knew nothing about its actual day-to-day tasks. So I got myself the Intro to Bioinformatics Algorithms book and trundled through it. This was an exceptionally boring and difficult reading, as "computational biology" turned out to be 95% string matching algorithms. Input: a DNA sequence and a bank of DNA records. Task: do a fuzzy search to identify both perfect and approximate (mutated) matches. Makes sense, but I couldn’t imagine myself innovating here.

I was better equipped to innovate in the field of AI, which I knew much better thanks to the AIMA book. I was doing some work in computational linguistics and game AI, and had some clear directions to pursue. Unfortunately for me, I could only achieve incremental improvements here and there, and while I still believe I could have pushed it harder, the real breakthroughs were obtained thanks to deep learning and all kinds of novel neural network architectures, which were not the directions I was interested in. Neural networks required massive hardware upgrades (which I couldn’t really afford) and far stronger mathematical skills than I possess. This way, my particular combination of talents and prior experience did not bring much value in the given historical circumstances.

Anyway, my point is simple: the general direction of the most ambitious future research was and is clear: biotech and AI. Now, I love all kinds of sciences, and I am not suggesting that we should or would slow down physics or linguistics, but in terms of transition to the "maturity phase" and potential impact I don’t really see other strong candidates. Physics is a "mature" science in a sense that it had a long way from pure observations to theories to engineering. We first observe what happens, then we describe what happens by building mathematical models, then we engineer what we want by applying our obtained knowledge. Going through observations and modeling is necessary and inevitable, but the real impact starts with engineering. It’s cool that Kepler or Newton could predict the movements of planets, but most normal people care about physics because it builds airplanes and nuclear stations. Right now, biology and medicine makes its early steps in engineering, and AI (engineering-oriented from day one) receives sufficient theoretical and technological support.

Other sciences will give us better cars or more powerful batteries (please do), but I don’t believe any of these changes would be truly transformative. If I am teleported back, say, to 1970s, I won’t go through too much pain to get used to their cars, airplanes, clothes, or furniture. In contrast, ideas like radical life extension or smart AI assistants are transformative if achieved. Transition to the engineering phase does not promise pies in the sky, but it makes at least some kinds of pies possible, and it’s hard to believe that we the humankind will suddenly decide not to pursue them. On top of that, the biotech/AI pies are an easy sell to the people with resources to fund research. There are numerous historic examples of anti-progress oriented elites, e.g., Tokugawa shogunate banning guns because they jeopardize the samurai class or 19-century Russian authorities wary of railroads because they force social classes to mix and help spread new disruptive ideas across the country. I don’t think that our emerging technologies pose any threat to the modern rulers, and some elements that are disruptive are easier to suppress than to abandon the whole research effort. If you are satisfied with your life, what’s your wish? To continue in the same manner as long as possible, which is a vastly superior alternative to getting old and sick, suffer, and die. AI can be seen as a pursuit on its own or as a way to multiply our mental abilities needed elsewhere (right, in biotech).

One may argue that there is nothing really new in seeking the elixir of life, but I think there is an important difference between someone like Qin Shi Huang and someone like Peter Thiel. For a pre-modern dreamer, to fly, to turn stone into gold, or to stay young was a matter of finding the right recipe "out there", by rummaging through ancient manuscripts or trying to receive divine revelation. For us, all these dreams pose a challenge to develop the right technology, and the only question is whether we already possess the expertise to attack the problem or not yet. In late 19th century people had enough knowledge to build a primitive airplane, but they lacked a critical part of the puzzle: a strong and lightweight engine able to lift the machine. As we know, steam engine wasn’t up to the task. Similarly, we know that our current tech is not up to the task of curing cancer, for example, but we do what we can do, and nobody thinks we are at the dead end. Rather opposite, there is a feeling that biology/medicine is about to enter its "engineering" phase. Obviously, the case of AI is different. By early 2000s there was a feeling that we are seemingly approaching a certain ceiling, but a combination of some fresh takes on old ideas, better hardware, and much available data did the trick: it’s hard to say where the new ceiling is now, but it definitely got a lot higher.

When thinking about all kinds of Big World Issues, I usually ask myself several things: 1) are commonly mentioned pain points at least theoretically fixable? 2) what is so different now comparing to similar situations in the past that makes people upset? 3) what is the ideal world the critics envision, and do I really want to live there?

I am obviously biased, but generally this framework generally leads me into the "pro-technology" camp, because usually I believe that most issues can be fixed, I don’t see much novelty in world events, and I don’t think the present world is somehow optimal to dread change. Yudkowsky is worried that superhuman AI "would kill us all", and I guess this is the kind of argument I can’t dismiss. (Personally I don’t believe in this scenario, but I do my best to separate my own opinion from realistic possibilities, however unlikely.) Arguments involving less dreadful scenarios often identify very real pain points, but they don’t make me lose my sleep.

For example, it is commonly pointed out that the modern AI systems require too much power, consume valuable water, drive prices, and are generally unsustainable. These are all valid concerns, but can’t they be fixed at least in theory? I am not qualified to judge that, but these seem to be purely technological challenges, requiring new power sources (e.g., nuclear), new locations (e.g., in cold regions), extra production volumes, etc. Furthermore, what’s new here? Yes, we hear a lot about how data centers damage environment and contribute to global warming aka climate change, but this is all being spoken as if without data centers we are any close to reverse the trend. Again, I am absolutely not qualified to judge, but it does not worry me much if data centers make some awful thing arrive 10 years earlier than we anticipated given that we still have absolutely no chance to prevent this awful thing according to all predictions. Maybe I am simply not old enough and believe I’ll live to see "the thing" in any case.

When people point out that AI helps "spread misinformation" or "foster propaganda", I honestly scratch my head — is the presumption here that we were not living in the world full of misinformation and propaganda? Each of us knows some stereotypical grandma who believed certain thing because "it was in the newspaper", and no kind of argument could have change her conviction. In a sense, I believe that "worse is better": we all were way too guileless and believed dubious information because it allegedly came from reputable source. Now we should learn to stay on guard and do not trust anything blindly, which is a good thing. Much of this "misinformation" narrative is a flat-out "us vs them" confrontation, where some comments are being labeled as supporting the wrong side, and nobody cares if the comment in question is factually accurate. I do want to live in a world where I can ponder on all kinds of narratives. It’s okay if this narrative is to the benefit of someone I don’t like; it’s actually the opposite — I do want to understand what kind of stories shape their worldview, and if I can get them only via bots, so be it.

Then what about scary stories about AI taking jobs and robbing people of their livelihoods, taking bread from the mouth of needy to the joy of the likes of Bezos or Zuckerberg? What about companies that use art and books to train their LLMs? Look, I am of course on the side of people who want to have meaningful occupation and wish to support themselves rather than being unemployed. I want the authors to get the deserved credit and receive what is due. We absolutely need to do something about them, but jamming the wheels of technological progress is not the answer. The technology itself does not automatically make people happy. Clearly, an average hunter-gatherer had a better life than a Victorian factory worker. And yet, what is the alternative here? Endless people doing mechanical, allegedly creative but actually not work like making generic ads, drawing generic characters and giving generic advices to people in generic situations? And if I am fine with using AI to draw a generic poster, am I a villain who robs someone from their livelihood? It’s not like I owe someone employment. We live in a society where bullshit jobs are abundant, and let’s not pretend that technology drains much of meaningful existence from us.

I am mostly talking about AI (I’ll set aside the question of whether AI is "the real intelligence" as purely theoretical) rather than biotech because we see some AI-induced impact already, while biotech is still only making some baby steps. However, the idea of radical life extension, purely speculative at the moment, ignites enough opposition. Will our planet sustain that? Will the rich upgrade themselves to live long and happy lives while the poor stay where they are now? Will dictators stay in power forever? I don’t dismiss these concerns, but personally I am not ready to endure pain and death only because someone else is poor or is suffering under dictatorships. I had my own share of misfortunes, and I should not be held responsible for the bad luck of other people. In any case, I don’t think this is a zero-sum game where I can get some extra years only by taking them from someone else.

The point of this lengthy discussion is merely to demonstrate that technological progress and disruptive technologies is old news. All the concerns we’ve been debating so far are as old as civilization: armed conflicts, worsening of life conditions, lack of jobs and money, inequality, and so on. And yet there is too much negativity around, at least, to my taste. This is somewhat surprising: are we the first generation that pours enormous resources into dubious projects to achieve elusive ends? Okay, I can’t get into the head of an average pyramid-building Egyptian, but at least we can recall the "rocket age" of 60s-70s.

I wasn’t there, so I can hardly speak with confidence, but judging from all kinds of sources it seems that the optimism was prevailing on both sides of the Iron Curtain. Popular science writers, science fiction writers, filmmakers and household item manufactures embraced rockets and space traveling with gusto. We had authors like Asimov and Bradbury, Star Wars and Star Trek, rocket-shaped vacuum cleaners, space themed music bands and comics. I am taking science fiction exploring the themes of space colonization and interstellar travel with a grain of salt: after all, the goal of a writer is to produce fiction that sells rather than to make accurate predictions. Fiction is not futurology. Yet these artists rode on the coattails of very real technical advancements and general public interest.

Why the general mood was so different back then? Frankly speaking, I don’t have a good answer. If you don’t like Elon Musk, do you like Wernher von Braun? That guy certainly did the Nazi salute many times. In his case it was not a one-time sloppy gesture caught on camera. If you think data centers is a waste of precious resources, what about Moon landing? The only purpose of that feat was to outdo the USSR, and people gladly applauded this goal. Nobody consulted Soviet people to choose priorities, but successful space expeditions were also widely cheered, and nobody in the position of power questioned the expenses. The Space Race certainly brought numerous useful byproducts — theories, technologies, industries, but I don’t really think one needs and artificial goal to make progress. Go cure cancer. Go automate menial soul-crushing jobs like garbage processing or fish gutting. That’s the challenge. If AI or biotech is not for the benefit of everyone, then Space Race isn’t either. Let’s keep in mind that the space program was initiated when the USA still had institutionalized segregation, and Soviet citizens had no right to move freely inside the country. No sci-fi technology did change that.

I tend to believe that today people became more sensitive to certain matters. They were okay with Lockheed Martin and Thiokol benefiting from the space program, but now they don’t like when Facebook or Google do the same. They did not care about environmental damage made by rocket launches, and now they are unhappy with data centers consuming electricity and water. They weren’t concerned with obvious military applications of rockets, and yet they worry that AI might be used for war and surveillance. Some might call our modern leaders untrustworthy, but men like Khrushchev and Nixon weren’t much better. We are living in times of disillusionment: instead of space travel and worldwide prosperity we have ads everywhere, AI-generated videos pretending to be genuine, depopulation, economic hardships, climate issues, and entirely unnecessary wars. Yet I think this is the case of disillusionment due to unreasonably rosy expectations: our present reality isn’t that dismal, but we had hoped for more.

As I mentioned before, we should not confuse fiction with futurology, but for some reason it seems that in the average person’s opinion space travels and Mars colonies were almost around the corner. Then the fashion turned from space operas to "high tech, low life" cyberpunk, and the current perception of progress is much less sanguine. The rich will get richer, the poor will get poorer, rising crime will be matched by mass surveillance, dictatorships will prosper, and our planet will turn into a mess. Classic cyberpunk novels liked rainy days for some reason. We’ll likely have heat waves instead.

This essay is getting overly lengthy, and I let my train of thought to take me way too far from the original question: why people complain about nearly the same things that made them happy in the past? Whatever the answer is, it won’t help us to feel better. My own approach is to use history as a reminder that everything depends on your point of view. The same thing may look great or abysmal, so if you believe something is definitely abysmal, think again. I am not a kind of person who embraces new things overly enthusiastically, it is quite the opposite. I was a very late adopter of smartphone, I never had bitcoins, I have very little presence in social media, and in general I prefer not to rush things. However, it is clear to me that the genie cannot be put back into his bottle. This ship has sailed. If technology breaks something, we should find the way to fix it with better technology rather than than to hit brakes. If one company or country hits breaks, another one would use it to its own benefit. We have to embrace this reality and take advantage of it.

I have no idea whether we are on the way to superhuman AI. Maybe we’ll hit a new ceiling soon. However, it is safe to say that the current agents are at least as skillful as a very good junior software developer. Whoever calls the resulting code 'slop' is welcome to review some graduation works. Do our students recognize that their junior skillset won’t land them a job anymore? Are they actively working to become better engineers, ready take the lead of AI juniors? Go guess, and don’t blame AI anymore. In a sense, my optimism has a tint of desperation: if we can’t adapt to this new reality, if we can’t fix our environment with technology, then there is no hope, and we are all doomed. It does not bother me too much that the greatest beneficiaries of progress will be unpleasant people and unpleasant corporations. There is nothing new about it. Our goal is to find out how to make our world a better place with the means we have or are likely to obtain soon. There are many ways to contribute: develop technology, advance theory, find practical applications, improve laws and regulations, educate. Luddism isn’t the option.