The Future of Sex

Our descendants will be different from us. In a competitive world, they’ll have to be; our design is hardly optimized for their world. But since they will evolve incrementally from us, they won’t be completely different.  For example, many features of the ways we talk between minds, and within minds, may lock in as interface standards.  Also, our descendants will prefer to reuse and modify complex workable modules rather than reinventing such things from scratch.

Which brings us to everyone’s favorite topic: sex. Our minds have been evolved in great detail to handle human sex. How might our descendants reuse and adapt those well-honed capabilities to deal with future mental challenges?

First, it is pretty obvious that within a century or two at most our descendants just won’t be creating descendants by randomly mixing the features of two parents, any more than firms today design new products via random mixes of old product features. No, our descendants will be more deliberately designed, with design components inspired by, if not directly taken from, a great many predecessors.  They just won’t make babies the bio-sex way.

Even so, our distant descendants will continue to form long-term alliances between minds whose qualities and loyalties are opaque. Even when one can directly peer inside, most complex minds simply have no clear place to look to see their overall abilities and loyalties. Such features are instead spread across such minds and best seen in actual behavior.  So to infer such features it can help to probe and test such minds in particular ways.  Our mental sexual toolkit is full of such ways to probe and test.

Also, when complex minds last longer than the multi-mind tasks they tackle, they must choose which minds combine to do which tasks.  And to create good incentives, minds must share some consequences of their joint performance, while committing in certain ways to outcomes they might not prefer after the fact.  Our sexual toolkit also has many useful ways to deal with these issues.

Our descendants will therefore likely recruit variations on our sexual toolkit to such tasks.   They will distinguish flings from “true love” while adapting human feelings of lust, romance, attachment, jealousy, and intimacy, and also variations on our mating dances of watching, displaying, flirting, wooing, testing, seducing, accusing, betraying, etc.

Our descendants may also distinguish male from female patterns of such behaviors. For example, some will pursue while others evaluate, some will take more risks while others play it safer, some will invest more vs. less in each relation, and some will protect against outside dangers while others nurture inside growth.

Our mental adaptations to sex are subtle and well-tuned for our mating task of slowly teasing out the abilities and intentions of others while becoming increasingly committed to and dependent on those others.  Our distant descendants will likely adapt such abilities for their many purposes.  Future sex may well change greatly to meet future needs, but it will still be recognizably sex all the same.  Long live sex!

GD Star Rating
Tagged as: , ,
Trackback URL:
  • Carl Shulman

    “First, it is pretty obvious that within a century or two at most our descendants just won’t be creating descendants by randomly mixing the features of two parents, any more than firms today design new products via random mixes of old product features.”

    Given this level of competitive and design pressure, it seems implausible that the minds would retain these human modules for very long at all. After millions of subjective years of work by the massive enhanced emulation population, still no non-emulation AI?

  • “but it will still be recognizably sex all the same”

    You might have different criteria for recognizing similarities than many people.

    In this post, you’ve highlighted the game-theoretic and design-work functions that sex performs, but you haven’t mentioned the subjective feelings that sex and romance invoke.

    In a scenario where software AIs “descended” from human uploads with no qualia iteratively examined each other’s software and verified that it satisfied certain commitment properties, you might say “they’re having sex”.

    Most people would say that the relevant criteria for comparison are: flirting, kissing, romance, human bodies and bodily contact, human genitalia, orgasm and feelings of romantic attachment and elation, none of which would be present in the Hansonian interaction, even though you are correct that it has “sex-like functionality”.

  • Carl, why would blank-slate designs have much of a chance against designs that build on and improve lots of inherited adaptations? Biology and modern engineering practice both rely heavily on incrementally improving prior designs, reusing workable modules when possible.

    Roko, I’m pretty confident ems have “qualia” and I’m suggesting that even without human genitalia it would still feel like sex, and that it would still include emotions, romance, flirting, elation, etc. I thought I was pretty clear on that point.

    • Blank slate designs might have an advantage over ems, because transistors are a different hardware substrate than neurons. Neurons are more plentiful, slower, and more error-prone. Neuron-based algorithms tend to do a lot of approximate pattern-matching.

      Computer chess programs reason very differently from human chess experts. (The ones that try to copy the detailed reasoning of human experts perform much worse.) Surely it’s reasonable to at least consider whether blank-slate AI might have significant design advantages over copies of brains.

      (Not to mention: since we don’t have access to the brain’s design documents, it might prove difficult to improve it much. Even if blank-slate AI starts out with worse performance, it seems likely that its growth in performance might maintain an edge over ems indefinitely.)

    • Tim Tyler

      Re: why would blank-slate designs have much of a chance against designs that build on and improve lots of inherited adaptations?

      The reason engineering from scratch has been so successful so far is because existing designs are unmaintainable spaghetti code, cobbled together by natural selection – and are difficult to rapidly modify and adapt to new purposes. The car was not an evolution of the horse, computers are not descended from brains, hydraulic lifters are not based on animal muscles, surveilance systems are not the distant offspring of guard dogs – and so on and so forth.

    • Tim, the quality of our mind code is a distinct issue from our current inability to directly modify it. We haven’t been able to directly modify our brain in meaningful ways, but will be more able to do that with ems. We only successfully manage complete redesigns of systems that are not locked in via current practice standards. Consider our failure to redesign human languages or education curricula or corporate business practices.

      • Carl Shulman

        In your competitive emulation scenarios, a new mind design that delivers a 10% productivity increase could support astronomical R&D development efforts, and rented hardware quickly reallocated to the new design, which could be copied up to the limits of hardware resources. In that scenario the benefits of shifting look huge relative to the costs.

      • Designers will focus on changes that do not also require a complete redesign of other larger entrenched standards and systems. Don’t see why that wouldn’t usually be possible.

      • Tim Tyler

        Takeovers are ubiquitous in evolution. I gave some examples already. There are more examples on my “Memetic Takeover” page:

        Automation provides many more examples. ATM machines are not a gradual evolution from bank clerks. Roombas are not a gradual evolution from human cleaners. Factory assembly robots are not a gradual evolution from assembly line workers.

        Engineering around lock-ins does not seem to be a big deal in such cases. You can rip out the human, stick in a machine, and it fits, works 24/7, needs no toilet breaks and never asks for a pay rise or goes on strike.

    • Tim Tyler

      Getting locked-in to sub-optimal standards is an way of describing one way of getting stuck on local maxima in design space. As our refactoring cababilities improve, we will be able to make larger leaps across design space to higher peaks – resulting in reduced lock-in effects.

      In this particular case, lock-in is does not seem very relevant – not much locks in our mind architecture. Even most of the mind’s I/O formats aren’t locked in – unless you want the machine mind to run a human body.

    • Tim, I don’t see that we are getting lots more able to avoid standards locks, and huge swaths of our society are adapted in detail to our mental styles.

      • Tim Tyler

        We do escape from many older standards, though. E.g. VHS.

        Our “mental styles” do not really lock in humans very effectively – since mental things can be – or will be able to be – reproduced by computers – speech recognition and synthesis, handwriting recognition and synthesis, machine intelligence, etc.

        Thus the waves of automation – where humans are replaced by machines.

    • > I’m pretty confident ems have “qualia”

      “Even if we do not suppose that uploading and outsourcing will result in a widespread loss of consciousness, we can still entertain the possibility that intrinsically valuable activities and states of consciousness become rarer or disappear altogether. Much of human life’s meaning arguably depends on the enjoyment, for its own sake, of humor, love, game-playing, art, sex, dancing, social conversation, philosophy, literature, scientific discovery, food and drink, friendship, parenting, and sport. We have preferences and capabilities that make us engage in such activities, and these predispositions were adaptive in our species’ evolutionary past; but what ground do we have for being confident that these or similar activities will continue to be adaptive in the future? Perhaps what will maximize fitness in the future will be nothing but non-stop high-intensity drudgery, work of a drab and repetitive nature, aimed at improving the eighth decimal of some economic output measure. Even if the workers selected for in this scenario were conscious, the resulting world would still be radically impoverished in terms of the qualities that give value to life”

      – Nick Bostrom

      • I don’t see how posting a quote of a long list of vague fears is responsive here.

      • Anonymous

        From what we currently know about the neuroscience of pleasure, it would be entirely possible for redesigned minds to strongly enjoy drudgery and repetitive work. Furthermore, “improving the eighth decimal of some economic output measure” could require a sophisticated skillset of complex and challenging intellectual activities, all of which could be intrinsically rewarding for adequately adapted minds.

  • Peter Twieg

    Even so, our distant descendants will continue to form long-term alliances between minds whose qualities and loyalties are opaque

    What makes you think that this will always be true? I could imagine future agents being able to credibly reveal their utility functions to one another, insofar as they’re able to disclose their utility functions at all (which I believe should be possible.) This might rely on reputational incentives, but I can’t imagine that cases where agents will meet without having access to each other’s reputations would be the very rare exception to the rule.

    I can also imagine future agents being able to fake loyalty (moreso than is already possible), especially if we’re simply relying on sex to reinforce coalitions internally rather than making commitments that are actually visible to all parties.

  • Peter, why would an agent’s “utility function” be a small localized thing, much smaller than the agent’s whole code? Why would faked commitments be easier in the future than today?

    • Tim Tyler

      Company mission statements are usually small things – much smaller than the whole company. Like that, I should think. It helps you when evaluating what you are doing. It helps when telling others what you are doing.

    • Corporations don’t hide their mission statements. Yes of course we’ll be able to find out what minds say is their utility function. The issue is what if we aren’t sure we believe them.

      • Tim Tyler

        I was addressing the “size” issue. There are some proposals for dealing with goal transparency – e.g. have the parties design an agent between them, and then send its bits to a fabricator – or use a “government approved” agent with a valid signature, and a visible utility function. Reputations may help too. With a record of an agent’s actions, you should often be able to see if its behaviour to date has been consistent with its stated aims. The longer the track record, the more trustworthy.

  • Aron

    Won’t it become rather immoral to haphazardly experiment with a sentient being’s future by creating novel mixtures of their attributes? I would suspect that most future adaptation eventually occurs at the level of a ‘responsible adult’ who makes the choice to self-modify. Cloning then becomes the norm where reproduction is necessary, usually with a specific means for income already in place (a constraint from regulatory agencies perhaps). It seems the post above presupposes a future where we have little ability to modify our own design in situ, yet we are talking of a future where most interesting stuff is digital.

  • Aron, of course individuals will modify themselves, but I expect that just can’t compete with purposeful periodic redesigns. Can’t see how those could be less moral than our current random combo approach.

    Don, chess programs are usually redesigned relatively incrementally. There is of course variation, and some changes are less incremental than others.

    • Aron

      I agree that our current model is even less moral than what you advocate. I would be arguing that our standards would change based on the fact that we now have a choice where before we did not. This kind of choice expansion underpins a lot of the evolution of morality.

      The alternative we’d be choosing between is to do ‘purposeful periodic redesigns’ on a being that is voluntarily selecting to do so, and one that is not.

  • Pingback: Tweets that mention Overcoming Bias : The Future of Sex --

  • Pingback: uberVU - social comments()

  • Tim Tyler

    I have a video entitled “Future sex” on this general topic:

    I concentrate on the issues of whether we will still have genders, whether they will arise in separate bodies (rather than being hermaphrodites), the number of ancestors that contribute to each offspring, and the “quantity” of recombination in each generation.

    • Violet

      Maybe we can just swap “bodies” and thus select whatever sexed bodies we prefer each time we have “sex”.

  • Will there still be alcohol before and cigarettes afterwards?

  • Philo

    “Future sex may well change greatly to meet future needs, but it will still be recognizably sex all the same. Long live sex!” This looks like a very shaky prediction, covering, as it does, the *whole future*. If our successors eventually abandon natural reproduction, preferring to engineer new lifeforms, who knows how far they may depart from traditional sex? Divorced from reproduction, sex will have lost its raison d’être, and will be ripe not just for modification but for replacement. Sure, this will be tricky, but our successors will have an indefinitely long time to work on it.

    • Tim Tyler

      Re: If our successors eventually abandon natural reproduction, preferring to engineer new lifeforms, who knows how far they may depart from traditional sex?

      As far as the information-theoretic aspects of sexual reproduction go, memeticists and computer programmers, I would propose. They deal with engineered heritable information on a day-to-day basis. The future may face *some* different issues – those associated with copyrights and patents, for example – but such folk will probably have the basics down.

  • David C

    Is the decision of who to mate with to raise a child really all that random today?

  • lemmy caution

    If our successors eventually abandon natural reproduction, preferring to engineer new lifeforms, who knows how far they may depart from traditional sex?

    People love sex and they love babies. Some people will give up natural reproduction, but not everyone without a substantial amount of coercion. Lots of people won’t worry about how their kids are going to compete with genetically modified super babies. They will just get pregnant and have kids like people have always done.

  • This doesn’t seem obvious. We have reliable and intimate allies today without having sex with them.

    Sex is only for the greatest commitment of sharing genes. Why do you think sex exists in the first place? The ‘red queen hypothesis’ was the most popular the last time I checked. Assuming it’s right, once we are uploads on fast hardware, it is unlikely that rapidly evolving parasites will present the same problem for us that they did to our ancestors because we will be able to back ourselves up and reprogram our software and hardware in anticipation to evade them.

    Why then not switch to asexuality and have many sex-less friends?

    • Tim Tyler

      The same way computer viruses are not a problem – because we can back our computers up and reprogram them – in anticipation of being attacked?

  • Rob, I’m saying we will reproduce sexually, I’m saying that we will not waste all the design effort that went into our sex mental toolkit. Yes, that toolkit is not the only way to manage relations, but it is a powerful way, and so we will make use of those tools.

  • Strange then that we don’t use sex already to scout out business partners and friends and such. Why haven’t we evolved to use sex modules for anything other than reproduction? Bonobos seem to have done so (though I hear that research is in question), but this is uncommon in the animal kingdom.

  • Rob, humans already do to some extent, though perhaps not as much as bonobos. I’m suggesting that our descendants will do more.

  • Pingback: Linkage is Good for You: Happy Fake Corporate Holiday Edition()