Tag Archives: Values

On Value Drift

The outcomes within any space-time region can be seen as resulting from 1) preferences of various actors able to influence the universe in that region, 2) absolute and relative power and influence of those actors, and 3) constraints imposed by the universe. Changes in outcomes across regions result from changes in these factors.

While you might mostly approve of changes resulting from changing constraints, you might worry more about changes due to changing values and influence. That is, you likely prefer to see more influence by values closer to yours. Unfortunately, the consistent historical trend has been for values to drift over time, increasing the distance between random future and current values. As this trend looks like a random walk, we see no obvious limit to how far values can drift. So if the value you place on the values of others falls rapidly enough with the distance between values, you should expect long term future values to be very wrong.

What influences value change?
Inertia – The more existing values are tied to important entrenched systems, the less they change.
Growth – On average, over time civilization collects more total influence over most everything.
Competition – If some values consistently win key competitive contests, those values become more common.
Influence Drift – Many processes that change the world produce random drift in agent influence.
Internal Drift – Some creatures, e.g., humans, have values that drift internally in complex ways.
Culture Drift – Some creatures, e.g., humans, have values that change together in complex ways.
Context – Many of the above processes depend on other factors, such as technology, wealth, a stable sun, etc.

For many of the above processes, rates of change are roughly proportional to overall social rates of change. As these rates of change have been increased over time, we should expect faster future change. Thus you should expect values to drift faster in the future than then did in the past, leading faster to wrong values. Also, people are living longer now than they did in the past. So even past people didn’t live long enough to see big enough changes to greatly bother them, future people may live to see much more change.

Most increases in the rates of change have been concentrated in a few sudden large jumps (associated with the culture, farmer, and industry transitions). As a result, you should expect that rates of change may soon increase greatly. Value drift may continue at past rates until it suddenly goes much faster.

Perhaps you discount the future rapidly, or perhaps the value you place on other values falls slowly with value distance. In these cases value drift may not disturb you much. Otherwise, the situation described above may seem pretty dire. Even if previous generations had to accept the near inevitability of value drift, you might not accept it now. You may be willing to reach for difficult and dangerous changes that could remake the whole situation. Such as perhaps a world government. Personally I see that move as too hard and dangerous for now, but I could understand if you disagree.

The people today who seem most concerned about value drift also seem to be especially concerned about humans or ems being replaced by other forms of artificial intelligence. Many such people are also concerned about a “foom” scenario of a large and sudden influence drift: one initially small computer system suddenly becomes able to grow far faster than the rest of the world put together, allowing it to quickly take over the world.

To me, foom seems unlikely: it posits an innovation that is extremely lumpy compared to historical experience, and in addition posits an unusually high difficulty of copying or complementing this innovation. Historically, innovation value has been distributed with a long thin tail: most realized value comes from many small innovations, but we sometimes see lumpier innovations. (Alpha Zero seems only weak evidence on the distribution of AI lumpiness.) The past history of growth rates increases suggests that within a few centuries we may see something, perhaps a very lumpy innovation, that causes a growth rate jump comparable in size to the largest jumps we’ve ever seen, such as at the origins of life, culture, farming, and industry. However, as over history the ease of copying and complementing such innovations has been increasing, it seems unlikely that copying and complementing will suddenly get much harder.

While foom seems unlikely, it does seems likely that within a few centuries we will develop machines that can outcompete biological humans for most all jobs. (Such machines might also outcompete ems for jobs, though that outcome is much less clear.) The ability to make such machines seems by itself sufficient to cause a growth rate increase comparable to the other largest historical jumps. Thus the next big jump in growth rates need not be associated with a very lumpy innovation. And in the most natural such scenarios, copying and complementing remain relatively easy.

However, while I expect machines that outcompete humans for jobs, I don’t see how that greatly increases the problem of value drift. Human cultural plasticity already ensures that humans are capable of expressing a very wide range of values. I see no obviously limits there. Genetic engineering will allow more changes to humans. Ems inherit human plasticity, and may add even more via direct brain modifications.

In principle, non-em-based artificial intelligence is capable of expressing the entire space of possible values. But in practice, in the shorter run, such AIs will take on social roles near humans, and roles that humans once occupied. This should force AIs to express pretty human-like values. As Steven Pinker says:

Artificial intelligence is like any other technology. It is developed incrementally, designed to satisfy multiple conditions, tested before it is implemented, and constantly tweaked for efficacy and safety.

If Pinker is right, the main AI risk mediated by AI values comes from AI value drift that happens after humans (or ems) no longer exercise such detailed frequent oversight.

It may be possible to create competitive AIs with protected values, i.e., so that parts where values are coded are small, modular, redundantly stored, and insulated from changes to the rest of the system. If so, such AIs may suffer much less from internal drift and cultural drift. Even so, the values of AIs with protected values should still drift due to influence drift and competition.

Thus I don’t see why people concerned with value drift should be especially focused on AI. Yes, AI may accompany faster change, and faster change can make value drift worse for people with intermediate discount rates. (Though it seems to me that altruistic discount rates should scale with actual rates of change, not with arbitrary external clocks.)

Yes, AI offers more prospects for protected values, and perhaps also for creating a world/universe government capable of preventing influence drift and competition. But in these cases if you are concerned about value drift, your real concerns are about rates of change and world government, not AI per se. Even the foom scenario just temporarily increases the rate of influence drift.

Your real problem is that you want long term stability in a universe that more naturally changes. Someday we may be able to coordinate to overrule the universe on this. But I doubt we are close enough to even consider that today. To quote a famous prayer:

God, grant me the serenity to accept the things I cannot change,
Courage to change the things I can,
And wisdom to know the difference.

For now value drift seems one of those possibly lamentable facts of life that we cannot change.

GD Star Rating
loading...
Tagged as: , ,

See A Wider View

Ross Douthat in the NYT:

From now on the great political battles will be fought between nationalists and internationalists, nativists and globalists. .. Well, maybe. But describing the division this way .. gives the elite side of the debate .. too much credit for being truly cosmopolitan.

Genuine cosmopolitanism is a rare thing. It requires comfort with real difference, with forms of life that are truly exotic relative to one’s own. .. The people who consider themselves “cosmopolitan” in today’s West, by contrast, are part of a meritocratic order that transforms difference into similarity, by plucking the best and brightest from everywhere and homogenizing them into the peculiar species that we call “global citizens.”

This species is racially diverse (within limits) and eager to assimilate the fun-seeming bits of foreign cultures — food, a touch of exotic spirituality. But no less than Brexit-voting Cornish villagers, our global citizens think and act as members of a tribe. They have their own distinctive worldview .. common educational experience, .. shared values and assumptions .. outgroups (evangelicals, Little Englanders) to fear, pity and despise. .. From London to Paris to New York, each Western “global city” .. is increasingly interchangeable, so that wherever the citizen of the world travels he already feels at home. ..

It is still possible to disappear into someone else’s culture, to leave the global-citizen bubble behind. But in my experience the people who do are exceptional or eccentric or natural outsiders to begin with .. It’s a problem that our tribe of self-styled cosmopolitans doesn’t see itself clearly as a tribe. .. They can’t see that paeans to multicultural openness can sound like self-serving cant coming from open-borders Londoners who love Afghan restaurants but would never live near an immigrant housing project.

You have values, and your culture has values. They are similar, and this isn’t a coincidence. Causation here mostly goes from culture to individual. And even if you did pick your culture, you have to admit that the young you who did was’t especially wise or well-informed. And you were unaware of many options. So you have to wonder if you’ve too easily accepted your culture’s values.

Of course your culture anticipates these doubts, and is ready with detailed stories on why your culture has the best values. Actually most stories you hear have that as a subtext. But you should wonder how well you can trust all this material.

Now, you might realize that for personal success and comfort, you have little to gain, and much to lose, by questioning your culture’s values. Your associates mostly share your culture, and are comforted more by your loyalty displays than your intellectual cleverness. Hey, everyone agrees cultures aren’t equal; someone has to be best. So why not give yours the benefit of the doubt? Isn’t that reasonable?

But if showing cleverness is really important to you, or if perhaps you really actually care about getting values right, then you should wonder what else you can do to check your culture’s value stories. And the obvious option is to immerse yourself in the lives and viewpoints of other cultures. Not just via the stories or trips your culture has set up to tell you of its superiority. But in ways that give those other cultures, and their members, a real chance. Not just slight variations on your culture, but big variations as well. Try to see a wider landscape of views, and then try to see the universe from many widely dispersed points on that landscape.

Yes if you are a big-city elite, try to see the world from Brexit or Trump fan views. But there are actually much bigger view differences out there. Try a islamic fundamentalist, or a Chinese nationalist. But even if you grow to be able to see the world as do most people in the world today, there still remain even bigger differences out there. Your distant ancestors were quite human, and yet they saw the universe very differently. Yes, they were wrong on some facts, but that hardly invalidates most of their views. Learn some ancient history, to see their views.

And if you already know some ancient history, perhaps the most alien culture you have yet to encounter is that of your human-like descendants. But we can’t possibly know anything about that yet, you say? I beg to differ. I introduce my new book with this meet-a-strange-culture rationale: Continue reading "See A Wider View" »

GD Star Rating
loading...
Tagged as: , ,