Most of you are probably familiar with the two contrasting decision making strategies "maximizing" and "satisficing", but a short recap won't hurt (you can skip the first two paragraphs if you get bored): Satisficing means selecting the first option that is good enough, i.e. that meets or exceeds a certain threshold of acceptability. In contrast, maximizing means the tendency to search for so long until the best possible option is found.

Research indicates (Schwartz et al., 2002) that there are individual differences with regard to these two decision making strategies. That is, some individuals – so called ‘maximizers’ – tend to extensively search for the optimal solution. Other people – ‘satisficers’ – settle for good enough¹. Satisficers, in contrast to maximizers, tend to accept the status quo and see no need to change their circumstances².

When the subject is raised, maximizing usually gets a bad rap. For example, Schwartz et al. (2002) found "negative correlations between maximization and happiness, optimism, self-esteem, and life satisfaction, and positive correlations between maximization and depression, perfectionism, and regret."

So should we all try to become satisficers? At least some scientists and the popular press seem to draw this conclusion:

Maximisers miss out on the psychological benefits of commitment, leaving them less satisfied than their more contented counterparts, the satisficers. ...Current research is trying to understand whether they can change. High-level maximisers certainly cause themselves a lot of grief.

I beg to differ. Satisficers may be more content with their lives, but most of us don't live for the sake of happiness alone. Of course, satisficing makes sense when not much is at stake³. However, maximizing also can prove beneficial, for the maximizers themselves and for the people around them, especially in the realm of knowledge, ethics, relationships and when it comes to more existential issues – as I will argue below⁴.

Belief systems and Epistemology

Ideal rationalists could be thought of as epistemic maximizers: They try to notice slight inconsistencies in their worldview, take ideas seriously, beware wishful thinking, compartmentalization, rationalizations, motivated reasoning, cognitive biases and other epistemic sins. Driven by curiosity, they don't try to confirm their prior beliefs, but wish to update them until they are maximally consistent and maximally correspondent with reality. To put it poetically, ideal rationalists as well as great scientists don't content themselves to wallow in the mire of ignorance but are imbued with the Faustian yearning to ultimately understand whatever holds the world together in its inmost folds.

In contrast, consider the epistemic habits of the average Joe Christian: He will certainly profess that having true beliefs is important to him. But he doesn't go to great lengths to actually make this happen. For example, he probably believes in an omnipotent and beneficial being that created our universe. Did he impartially weigh all available evidence to reach this conclusion? Probably not. More likely is that he merely shares the beliefs of his parents and his peers. However, isn't he bothered by the problem of evil or Occam's razor? And what about all those other religions whose adherents believe with the same certainty in different doctrines?

Many people don’t have good answers to these questions. Their model of how the world works is neither very coherent nor accurate but it's comforting and good enough. They see little need to fill the epistemic gaps and inconsistencies in their worldview or to search for a better alternative. Thus, one could view them as epistemic satisficers. Of course, all of us exhibit this sort of epistemic laziness from time to time. In the words of Jonathan Haidt (2013):

We take a position, look for evidence that supports it, and if we find some evidence—enough so that our position “makes sense”—we stop thinking.

Usually, I try to avoid taking cheap shots at religion and therefore I want to note that similar points apply to many non-theistic belief systems.

Ethics

Let's go back to average Joe: he presumably obeys the dictates of the law and his religion and occasionally donates to (ineffective) charities. Joe probably thinks that he is a “good” person and many people would likely agree. This leads us to an interesting question: how do we typically judge the morality of our own actions?

Let's delve into the academic literature and see what it has to offer: In one exemplary study, Sachdeva et al. (2009) asked participants to write a story about themselves using either morally positive words (e.g. fair, nice) or morally negative words (e.g. selfish, mean). Afterwards, the participants were asked if and how much they would like to donate to a charity of their choice. The result: Participants who wrote a story containing the positive words donated only one fifth as much as those who wrote a story with negative words.

This effect is commonly referred to as moral licensing: People with a recently boosted moral self-concept feel like they have done enough and see no need to improve the world even further. Or, as McGonigal (2011) puts it (emphasis mine):

When it comes to right and wrong, most of us are not striving for moral perfection. We just want to feel good enough – which then gives us permission to do whatever we want.

Another well known phenomenon is scope neglect. One explanation for scope neglect is the "purchase of moral satisfaction" proposed by Kahneman and Knetsch (1992): Most people don't try to do as much good as possible with their money, they only spend just enough cash to create a "warm-fuzzy feeling" in themselves.

Phenomenons like "moral licensing" and "purchase of moral satisfaction" indicate that it is all too human to only act as altruistic as is necessary to feel or seem good enough. This could be described as "ethical satisficing" because people just follow the course of action that meets or exceeds a certain threshold of moral goodness. They don't try to carry out the morally optimal action or an approximation thereof (as measured by their own axiology).

I think I cited enough academic papers in the last paragraphs so let's get more speculative: Many, if not most people⁵ tend to be intuitive deontologists⁶. Deontology basically posits that some actions are morally required, and some actions are morally forbidden. As long as you do perform the morally required ones and don't engage in morally wrong actions you are off the hook. There is no need to do more, no need to perform supererogatory acts. Not neglecting your duties is good enough. In short, deontology could also be viewed as ethical satisficing (see footnote 7 for further elaboration).

In contrast, consider deontology's arch-enemy: Utilitarianism. Almost all branches of utilitarianism share the same principal idea: That one should maximize something for as many entities as possible. Thus, utilitarianism could be thought of as ethical maximizing⁸.

Effective altruists are an even better example for ethical maximizers because they actually try to identify and implement (or at least pretend to try) the most effective approaches to improve the world. Some conduct in-depth research and compare the effectiveness of hundreds of different charities to find the ones that save the most lives with as little money as possible. And rumor has it there are people who have even weirder ideas about how to ethically optimize literally everything. But more on this later.

Friendships and conversations

Humans intuitively assume that the desires and needs of other people are similar to their own ones. Consequently, I thought that everyone secretly yearns to find like-minded companions with whom one can talk about one’s biggest hopes as well as one’s greatest fears and form deep, lasting friendships.

But experience tells me that I was probably wrong, at least to some degree: I found it quite difficult to have these sorts of conversations with a certain kind of people, especially in groups (luckily, I’ve found also enough exceptions). It seems that some people are satisfied as long as their conversations meet a certain, not very high threshold of acceptability. Similar observations could be made about their friendships in general. One could call them social or conversational satisficers. By the way, this time research actually suggests that conversational maximizing is probably better for your happiness than small talk (Mehl et al., 2008).

Interestingly, what could be called "pluralistic superficiality" may account for many instances of small talk and superficial friendships since everyone experiences this atmosphere of boring triviality but thinks that the others seem to enjoy the conversations. So everyone is careful not to voice their yearning for a more profound conversation, not realizing that the others are suppressing similar desires.

Crucial Considerations and the Big Picture

On to the last section of this essay. It’s even more speculative and half-baked than the previous ones, but it may be the most interesting, so bear with me.

Research suggests that many people don’t even bother to search for answers to the big questions of existence. For example, in a representative sample of 603 Germans, 35% of the participants could be classified as existentially indifferent, that is they neither think their lives are meaningful nor suffer from this lack of meaning (T. Schnell, 2008).

The existential thirst of the remaining 65% is presumably harder to satisfy, but how much harder? Many people don't invest much time or cognitive resources in order to ascertain their actual terminal values and how to optimally reach them – which is arguably of the utmost importance. Instead they appear to follow a mental checklist containing common life goals (one could call them "cached goals") such as a nice job, a romantic partner, a house and probably kids. I’m not saying that such goals are “bad” – I also prefer having a job to sleeping under the bridge and having a partner to being alone. But people usually acquire and pursue their (life) goals unsystematically and without much reflection which makes it unlikely that such goals exhaustively reflect their idealized preferences. Unfortunately, many humans are so occupied by the pursuit of such goals that they are forced to abandon further contemplation of the big picture.

Furthermore, many of them lack the financial, intellectual or psychological capacities to ponder complex existential questions. I'm not blaming subsistence farmers in Bangladesh for not reading more about philosophy, rationality or the far future. But there are more than enough affluent, highly intelligent and inquisitive people who certainly would be able to reflect about crucial considerations. Instead, they spend most of their waking hours maximizing nothing but the money in their bank accounts or interpreting the poems of some arabic guy from the 7th century⁹.

Generally, many people seem to take the current rules of our existence for granted and content themselves with the fundamental evils of the human condition such as aging, needless suffering or death. Whatever the reason may be, they don't try to radically change the rules of life and their everyday behavior seems to indicate that they’ve (gladly?) accepted their current existence and the human condition in general. One could call them existential satisficers.

Contrast this with the mindset of transhumanism. Generally, transhumanists are not willing to accept the horrors of nature and realize that human nature itself is deeply flawed. Thus, transhumanists want to fundamentally alter the human condition and aim to eradicate, for example, aging, unnecessary suffering and ultimately death. Through various technologies transhumanists desire to create an utopia for everyone. Thus, transhumanism could be thought of as existential maximizing¹⁰.

However, existential maximizing and transhumanism are not very popular. Quite the opposite, existential satisficing – accepting the seemingly unalterable human condition – has a long philosophical tradition. To give some examples: The otherwise admirable Stoics believed that the whole universe is pervaded and animated by divine reason. Consequently, one should cultivate apatheia and calmly accept one's fate. Leibniz even argued that we already live in the best of all possible worlds. The mindset of existential satisficing can also be found in Epicureanism and arguably in Buddhism. Lastly, religions like Christianity or Islam are generally against transhumanism, partly because this amounts to “playing God”. Which is understandable from their point of view because why bother fundamentally transforming the human condition if everything will be perfect in heaven anyway?

One has to grant ancient philosophers that they couldn't even imagine that one day humanity would acquire the technological means to fundamentally alter the human condition. Thus it is no wonder that Epicurus argued that death is not to be feared or that the Stoics believed that disease or poverty are not really bad: It is all too human to invent rationalizations for the desirability of actually undesirable, but (seemingly) inevitable things – be it death or the human condition itself.

But many contemporary intellectuals can't be given the benefit of the doubt. They argue explicitly against trying to change the human condition. To name a few: Bernard Williams believed that death gives life meaning. Francis Fukuyama called transhumanism the world's most dangerous idea. And even Richard Dawkins thinks that the fear of death is "whining" and that the desire for immortality is "presumptuous"¹¹:

Be thankful that you have a life, and forsake your vain and presumptuous desire for a second one.

With all that said, "run-off-the-mill" transhumanism arguably still doesn't go far enough. There are at least two problems I can see: 1) Without a benevolent superintelligent singleton "Moloch" (to use Scott Alexander's excellent wording) will never be defeated. 2) We are still uncertain about ontology, decision theory, epistemology and our own terminal values. Consequently, we need some kind of process which can help us to understand those things or we will probably fail to rearrange reality until it conforms with our idealized preferences.

Therefore, it could be argued that the ultimate goal is the creation of a benevolent superintelligence or Friendly AI (FAI) whose values are aligned with ours. There are of course numerous objections to the whole superintelligence strategy in general and to FAI in particular, but I won’t go into detail here because this essay is already too long.

Nevertheless – however unlikely – it seems possible that with the help of a benevolent superintelligence we could abolish all gratuitous suffering and achieve an optimal mode of existence. We could become posthuman beings with god-like intellects, our ecstasy outshining the surrounding stars, and transforming the universe until one happy day all wounds are healed, all despair dispelled and every (idealized) desire fulfilled. To many this seems like sentimental and wishful eschatological speculation but for me it amounts to ultimate existential maximizing^{12, 13}.

Conclusion

The previous paragraphs shouldn’t fool one into believing that maximizing has no serious disadvantages. The desire to aim higher, become stronger and to always behave in an optimally goal-tracking way can easily result in psychological overload and subsequent surrender. Furthermore, it seems that adopting the mindset of a maximizer increases the tendency to engage in upward social comparisons and counterfactual thinking which contribute to depression as research has shown.

Moreover, there is much to be learnt from stoicism and satisficing in general: Life isn't always perfect and there are things one cannot change; one should accept one's shortcomings – if they are indeed unalterable; one should make the best of one's circumstances. In conclusion, better be a happy satisficer whose moderate productivity is sustainable than be a stressed maximizer who burns out after one year. See also these two essays which make similar points.

All that being said, I still favor maximizing over satisficing. If our ancestors had all been satisficers we would still be picking lice off each other’s backs¹⁴. And only by means of existential maximizing can we hope to abolish the aforementioned existential evils and all needless suffering – even if the chances seem slim.

[Originally posted a longer, more personal version of this essay on my own blog]

Footnotes

[1] Obviously this is not a categorical classification, but a dimensional one.

[2] To put it more formally: The utility function of the ultimate satisficer would assign the same (positive) number to each possible world, i.e. the ultimate satisficer would be satisfied with every possible world. The less possible worlds you are satisfied with (i.e. the higher your threshold of acceptability), the less possible worlds exist between which you are indifferent, the less of a satisficer and the more of a maximizer you are. Also note: Satisficing is not irrational in itself. Furthermore, I’m talking about the somewhat messy psychological characteristics and (revealed) preferences of human satisficers/maximizers. Read these posts if you want to know more about satisficing vs. maximizing with regard to AIs.

[3] Rational maximizers take the value of information and opportunity costs into account.

[4] Instead of "maximizer" I could also have used the term "optimizer".

[5] E.g. in the "Fat Man" version of the famous trolley dilemma, something like 90% of subjects don't push a fat man onto the track, in order to save 5 other people. Also, utilitarians like Peter Singer don't exactly get rave reviews from most folks. Although there is some conflicting research (Johansson-Stenman, 2012). Furthermore, the deontology vs. utilitarianism distinction itself is limited. See e.g. "The Righteous Mind" by Jonathan Haidt.

[6] Of course, most people are not strict deontologists. They are also intuitive virtue ethicists and care about the consequences of their actions.

[7] Admittedly, one could argue that certain versions of deontology are about maximally not violating certain rules and thus could be viewed as ethical maximizing. However, in the space of all possible moral actions there exist many actions between which a deontologist is indifferent, namely all those actions that exceed the threshold of moral acceptability (i.e. those actions that are not violating any deontological rule). To illustrate this with an example: Visiting a friend and comforting him for 4 hours or using the same time to work and subsequently donating the earned money to a charity are both morally equivalent from the perspective of (many) deontological theories – as long as one doesn’t violate any deontological rule in the process. We can see that this parallels satisficing.

Contrast this with (classical) utilitarianism: In the space of all possible moral actions there is only one optimal moral action for an utilitarian and all other actions are morally worse. An (ideal) utilitarian searches for and implements the optimal moral action (or tries to approximate it because in real life one is basically never able to identify, let alone carry out the optimal moral action). This amounts to maximizing. Interestingly, this inherent demandingness has often been put forward as a critique of utilitarianism (and other sorts of consequentialism) and satisficing consequentialism has been proposed as a solution (Slote, 1984). Further evidence for the claim that maximizing is generally viewed with suspicion.

[8] The obligatory word of caution here: following utilitarianism to the letter can be self-defeating if done in a naive way.

[9] Nick Bostrom (2014) expresses this point somewhat harshly:

A colleague of mine likes to point out that a Fields Medal (the highest honor in mathematics) indicates two things about the recipient: that he was capable of accomplishing something important, and that he didn't.

As a general point: Too many people end up as money-, academia-, career- or status-maximizers although those things often don’t reflect their (idealized) preferences.

[10] Of course there are lots of utopian movements like socialism, communism or the Zeitgeist movement. But all those movements make the fundamental mistake of ignoring or at least heavily underestimating the importance of human nature. Creating utopia merely through social means is impossible because most of us are, by our very nature, too selfish, status-obsessed and hypocritical and cultural indoctrination can hardly change this. To deny this, is to simply misunderstand the process of natural selection and evolutionary psychology. Secondly, even if a socialist utopia were to come true, there still would exist unrequited love, disease, depression and of course death. To abolish those things one has to radically transform the human condition itself.

[11] Here is another quote:

We are going to die, and that makes us the lucky ones. Most people are never going to die because they are never going to be born. [….] We privileged few, who won the lottery of birth against all odds, how dare we whine at our inevitable return to that prior state from which the vast majority have never stirred?

― Richard Dawkins in "Unweaving the Rainbow"

[12] It’s probably no coincidence that Yudkowsky named his blog "Optimize Literally Everything" which adequately encapsulates the sentiment I tried to express here.

[13] Those interested in or skeptical of the prospect of superintelligent AI, I refer to "Superintelligence: Paths, Dangers and Strategies" by Nick Bostrom.

[14] I stole this line from Bostrom’s “In Defense of Posthuman Dignity”.

References

Bostrom, N. (2014). Superintelligence: Paths, dangers, strategies. Oxford University Press.

Haidt, J. (2013). The righteous mind: Why good people are divided by politics and religion. Random House LLC.

Johansson-Stenman, O. (2012). Are most people consequentialists? Economics Letters, 115 (2), 225-228.

Kahneman, D., & Knetsch, J. L. (1992). Valuing public goods: the purchase of moral satisfaction. Journal of environmental economics and management, 22(1), 57-70.

McGonigal, K. (2011). The Willpower Instinct: How Self-Control Works, Why It Matters, and What You Can Do to Get More of It. Penguin.

Mehl, M. R., Vazire, S., Holleran, S. E., & Clark, C. S. (2010). Eavesdropping on Happiness Well-Being Is Related to Having Less Small Talk and More Substantive Conversations. Psychological Science, 21(4), 539-541.

Sachdeva, S., Iliev, R., & Medin, D. L. (2009). Sinning saints and saintly sinners the paradox of moral self-regulation. Psychological science, 20(4), 523-528.

Schnell, T. (2010). Existential indifference: Another quality of meaning in life. Journal of Humanistic Psychology, 50(3), 351-373.

Schwartz, B. (2000). Self determination: The tyranny of freedom. American Psychologist, 55, 79–88.

Schwartz, B., Ward, A., Monterosso, J., Lyubomirsky, S., White, K., & Lehman, D. R. (2002). Maximizing versus satisficing: happiness is a matter of choice. Journal of personality and social psychology, 83(5), 1178.

Slote, M. (1984). “Satisficing Consequentialism”. Proceedings of the Aristotelian Society, 58: 139–63.

AI risk

Bullet points

• By all indications, an Artificial Intelligence could someday exceed human intelligence.
• Such an AI would likely become extremely intelligent, and thus extremely powerful.
• Most AI motivations and goals become dangerous when the AI becomes powerful.
• It is very challenging to program an AI with fully safe goals, and an intelligent AI would likely not interpret ambiguous goals in a safe way.
• A dangerous AI would be motivated to seem safe in any controlled training setting.
• Not enough effort is currently being put into designing safe AIs.

 

Executive summary

The risks from artificial intelligence (AI) in no way resemble the popular image of the Terminator. That fictional mechanical monster is distinguished by many features – strength, armour, implacability, indestructability – but extreme intelligence isn’t one of them. And it is precisely extreme intelligence that would give an AI its power, and hence make it dangerous.

The human brain is not much bigger than that of a chimpanzee. And yet those extra neurons account for the difference of outcomes between the two species: between a population of a few hundred thousand and basic wooden tools, versus a population of several billion and heavy industry. The human brain has allowed us to spread across the surface of the world, land on the moon, develop nuclear weapons, and coordinate to form effective groups with millions of members. It has granted us such power over the natural world that the survival of many other species is no longer determined by their own efforts, but by preservation decisions made by humans.

In the last sixty years, human intelligence has been further augmented by automation: by computers and programmes of steadily increasing ability. These have taken over tasks formerly performed by the human brain, from multiplication through weather modelling to driving cars. The powers and abilities of our species have increased steadily as computers have extended our intelligence in this way. There are great uncertainties over the timeline, but future AIs could reach human intelligence and beyond. If so, should we expect their power to follow the same trend? When the AI’s intelligence is as beyond us as we are beyond chimpanzees, would it dominate us as thoroughly as we dominate the great apes?

There are more direct reasons to suspect that a true AI would be both smart and powerful. When computers gain the ability to perform tasks at the human level, they tend to very quickly become much better than us. No-one today would think it sensible to pit the best human mind again a cheap pocket calculator in a contest of long division. Human versus computer chess matches ceased to be interesting a decade ago. Computers bring relentless focus, patience, processing speed, and memory: once their software becomes advanced enough to compete equally with humans, these features often ensure that they swiftly become much better than any human, with increasing computer power further widening the gap.

The AI could also make use of its unique, non-human architecture. If it existed as pure software, it could copy itself many times, training each copy at accelerated computer speed, and network those copies together (creating a kind of “super-committee” of the AI equivalents of, say, Edison, Bill Clinton, Plato, Einstein, Caesar, Spielberg, Ford, Steve Jobs, Buddha, Napoleon and other humans superlative in their respective skill-sets). It could continue copying itself without limit, creating millions or billions of copies, if it needed large numbers of brains to brute-force a solution to any particular problem.

Our society is setup to magnify the potential of such an entity, providing many routes to great power. If it could predict the stock market efficiently, it could accumulate vast wealth. If it was efficient at advice and social manipulation, it could create a personal assistant for every human being, manipulating the planet one human at a time. It could also replace almost every worker in the service sector. If it was efficient at running economies, it could offer its services doing so, gradually making us completely dependent on it. If it was skilled at hacking, it could take over most of the world’s computers and copy itself into them, using them to continue further hacking and computer takeover (and, incidentally, making itself almost impossible to destroy). The paths from AI intelligence to great AI power are many and varied, and it isn’t hard to imagine new ones.

Of course, simply because an AI could be extremely powerful, does not mean that it need be dangerous: its goals need not be negative. But most goals become dangerous when an AI becomes powerful. Consider a spam filter that became intelligent. Its task is to cut down on the number of spam messages that people receive. With great power, one solution to this requirement is to arrange to have all spammers killed. Or to shut down the internet. Or to have everyone killed. Or imagine an AI dedicated to increasing human happiness, as measured by the results of surveys, or by some biochemical marker in their brain. The most efficient way of doing this is to publicly execute anyone who marks themselves as unhappy on their survey, or to forcibly inject everyone with that biochemical marker.

This is a general feature of AI motivations: goals that seem safe for a weak or controlled AI, can lead to extremely pathological behaviour if the AI becomes powerful. As the AI gains in power, it becomes more and more important that its goals be fully compatible with human flourishing, or the AI could enact a pathological solution rather than one that we intended. Humans don’t expect this kind of behaviour, because our goals include a lot of implicit information, and we take “filter out the spam” to include “and don’t kill everyone in the world”, without having to articulate it. But the AI might be an extremely alien mind: we cannot anthropomorphise it, or expect it to interpret things the way we would. We have to articulate all the implicit limitations. Which may mean coming up with a solution to, say, human value and flourishing – a task philosophers have been failing at for millennia – and cast it unambiguously and without error into computer code.

Note that the AI may have a perfect understanding that when we programmed in “filter out the spam”, we implicitly meant “don’t kill everyone in the world”. But the AI has no motivation to go along with the spirit of the law: its goals are the letter only, the bit we actually programmed into it. Another worrying feature is that the AI would be motivated to hide its pathological tendencies as long as it is weak, and assure us that all was well, through anything it says or does. This is because it will never be able to achieve its goals if it is turned off, so it must lie and play nice to get anywhere. Only when we can no longer control it, would it be willing to act openly on its true goals – we can but hope these turn out safe.

It is not certain that AIs could become so powerful, nor is it certain that a powerful AI would become dangerous. Nevertheless, the probabilities of both are high enough that the risk cannot be dismissed. The main focus of AI research today is creating an AI; much more work needs to be done on creating it safely. Some are already working on this problem (such as the Future of Humanity Institute and the Machine Intelligence Research Institute), but a lot remains to be done, both at the design and at the policy level.

MIRI recently published "Smarter than Us", a 50 page booklet laying out the case for considering AI as an existential risk. But many people have asked for a shorter summary, to be handed out to journalists for example. So I put together the following 2-page text, and would like your opinion on it.

In this post, I'm not so much looking for comments along the lines of "your arguments are wrong", but more "this is an incorrect summary of MIRI/FHI's position" or "your rhetoric is infective here".

AI risk

Bullet points

• The risks of artificial intelligence are strongly tied with the AI’s intelligence.
• There are reasons to suspect a true AI could become extremely smart and powerful.
• Most AI motivations and goals become dangerous when the AI becomes powerful.
• It is very challenging to program an AI with safe motivations.
• Mere intelligence is not a guarantee of safe interpretation of its goals.
• A dangerous AI will be motivated to seem safe in any controlled training setting.
• Not enough effort is currently being put into designing safe AIs.

Executive summary

The risks from artificial intelligence (AI) in no way resemble the popular image of the Terminator. That fictional mechanical monster is distinguished by many features – strength, armour, implacability, indestructability – but extreme intelligence isn’t one of them. And it is precisely extreme intelligence that would give an AI its power, and hence make it dangerous.

In That alien message, Eliezer made some pretty wild claims:

My moral - that even Einstein did not come within a million light-years of making efficient use of sensory data.

Riemann invented his geometries before Einstein had a use for them; the physics of our universe is not that complicated in an absolute sense.  A Bayesian superintelligence, hooked up to a webcam, would invent General Relativity as a hypothesis - perhaps not the dominant hypothesis, compared to Newtonian mechanics, but still a hypothesis under direct consideration - by the time it had seen the third frame of a falling apple.  It might guess it from the first frame, if it saw the statics of a bent blade of grass.

They never suspected a thing.  They weren't very smart, you see, even before taking into account their slower rate of time.  Their primitive equivalents of rationalists went around saying things like, "There's a bound to how much information you can extract from sensory data."  And they never quite realized what it meant, that we were smarter than them, and thought faster.

In the comments, Will Pearson asked for "some form of proof of concept". It seems that researchers at Cornell - Schmidt and Lipson - have done exactly that. See their video on Guardian Science:

'Eureka machine' can discover laws of nature - The machine formulates laws by observing the world and detecting patterns in the vast quantities of data it has collected