Singularity Summit - Day 2
Another great day at the Summit. I think I have enough new ideas swimming in my head to blog about for a year, although I don’t want to turn this into an all-Singularity-all-the-time sort of thing. I’d like to give my overall impressions of the summit here and let the rest sink in a bit more before I try to post my own ideas on the Singularity. Please keep in mind that this summit was pretty much my crash course in all things AI.
To me, the biggest issues that were raised at the summit were:
1) The Singularity as the future of all human progress.
2) The societal reactions to the coming Singularity and technological advancement in general.
3) Moral/ethical questions involved in the creation of sentient beings.
During the conference many fields were discussed: biology, physics, nanotech, philosophy, finance etc. Essentially, all fields of study, all of human progress is leading up to the Singularity. Think of it this way: the Walkman was a predecessor of the iPod, the record player was a predecessor of the Walkman, the phonograph was a predecessor of the record player. All technology will be the precursor to the Singularity. From the first stone tool, technology began to diversify. Soon we had the wheel, agriculture, medicine, etc. In recent years I believe we have seen a shift from a divergence of specialized technologies, to a convergence of those parts into greater, cumulative technologies. Laser eye surgery is a great example of seemingly different fields, combining to great success. The Singularity will be the ultimate convergence of human progress.
That is, if we ever get there. The cultural and political hurdles to the Singularity are enormous. People fear new technology, especially technologies that could be potentially dangerous. With the uproars over genetically modified food, it’s hard to imagine that people will be ready to accept something like mind uploading. Even more distressing is government’s tendency to hijack or crush new advancements. One thing I found disturbing was how much of AI and robotic research is funded by the military. Do we really want to create super-intelligent killing machines? In my mind, the greatest obstacle we face as a species is overcoming governmental and cultural pressures that hold back or corrupt innovation.
Finally, there are the moral/ethical issues surrounding the creation of super-intelligent AI’s. There was a lot of speculation as to how we could instill some sort of conscience into machines that will be vastly more intelligent than we are. Even if we are able to give them some sort of morality, whose morality will it be? Eliezer Yudkowsy suggested today that as our morals are constantly evolving (think about race relations in America over the last 200 years), we should somehow give the AI’s that trajectory of moral growth, rather than a fixed set of rules that we think of as “good” today. I thought this was a very interesting idea. Beyond creating moral AI’s, there is the question of our own behavior towards AI’s. There was a lot of disagreement on this point at the conference. Ideas ranged from programming the AI’s to be happy as our slaves to treating them just like a human. Interesting stuff.
Overall, I would say that the Singularity is very exciting and very promising. The conference really motivated me to keep learning more about this, which I hope to share with you.
Here are some links to speakers and subjects that I found particularly interesting:
Accelarating Studies Foundation
Christine Peterson of the Foresight Nanotech Institute
And in conclusion, a great quote from Archimedes (I’m not sure whose presentation it was in, apologies.):
Give me a place to stand and I will move the world.
I don’t think the tendency for military research budgets to push the boundaries of technology outward is going to abate any time soon. The near future features, I fear, armies that are composed mostly of killer robots, remote controlled from the homeland. We already have the beginning of this, with predators and, more recently, SWORDs. Soldiers will just be guys sitting back in easy chairs, playing computer games that kill people.
On a different note, I come at the issue of instilling friendlies in AI from a bit of a contrary perspective. I don’t think it will be necessary, because whatever posthuman entity comes next won’t be a pure descendant of AI. Let me put it this way: the internet has already become the manifest collective subconcious of our species, a sort of shared mental prosthesis that has been indelibly marked by our influence (see: 6 million cat pictures, god knows how many billions of female anatomy). Now, the first AI is almost certain to be a creature of the internet. It may even become the internet, or the internet may become it (for instance, if one of the Singularity Institute’s seed AIs gets loose). If growing directly out of the shared experience of humanity doesn’t produce an entity that would prefer to keep us around out of some sort of filial piety, we don’t deserve to survive as a species.
That’s a great point, Matt. I think that you’re right about AI and the internet being heavily, if not completely, intertwined. Hopefully, the AI will be able to experience humanity outside of the internet as well. The last thing we need is a super-intelligent porn addict.
A few speakers at the Singularity Summit touched on the need for humanity to continue to improve itself in preparation for the Singularity. If we had an evolutionary type of AI that learned like a human child, we would presumably want to be the best parents possible. In my mind that would mean treating other people and animals with kindness in hopes that the AI would adopt those characteristics. I think this idea also applies to alien life. If aliens were to observe humans treating each other terribly, they might conclude that we are not a very nice species and should be eliminated before we colonize other planets. Who knows?
I think that whether the future holds AI, aliens, the second coming of Christ or any number of things we can’t even conceive of, the more loving and content we are individually, the better everything will turn out. We have a much better chance of a friendly Singularity if we have a friendly world to welcome it into.
I think mind uploading would be amazing. Even if it was just for personal use (like a yearbook for that point in my life).
I think the creation of applications that combine social networks is a step toward singularity and AI. It makes the entire internet available in one window.