Shopping for listeners?
I agree with his pessimism about space. I think if we ever live in space in large numbers, it will mostly be in something like O'Neill Cylinder colonies; and that this will be decades away.
I also agree with him about Skynet scenarios not being much of a worry; however, I see no problem with some people devoting time and attention to the issue -- so long as they don't distract the public discourse about the good that AI will bring. If there is a small risk of billions dying from rogue AI, then it's worth devoting some amount of resources to thinking about.
Genetics will impact society, definitely; but I wouldn't count on it happening that quickly. The biggest impacts people will probably see personally are things like illegal immigrants being forced to give up their DNA and better screening for diseases. Fetuses may get scanned for genetic diseases that run in the family. I don't think most people -- at least not for decades -- will get gene therapy for illnesses.
One area that I think will grow explosively in the coming years is Brain Computer Interfaces (BCIs). Not the ones based on EEGs, which are low-quality signals -- I'm talking about the kind that Mark Chevillet is working on at Facebook (non-invasive), Mary Lou Jepsen is working on at Openwater (non-invasive), and the invasive kinds being worked on at Kernel and Neuralink. In addition, there are many new, cheap probes being built for animal experiments (e.g. Neuralpixel probes).
There are several ways that BCIs and neural probes will impact society. The Google Brain researcher (and Stanford professor) David Sussillo recently wrote a tweet about the coming explosion in neural data, and how it will impact Systems Neuroscience:
In summary, systems neuroscience is about get bonkers. The advent of inexpensive, high-quality recording equipment with the maturation of deep learning is an incredible opportunity!
Couldn't be a better time to be in the field, but we should be preparing now.
And he gave a talk recently at the Simons Institute on how to use large amounts of spike data from neural probes, in combination with Deep Learning, to model the dynamics in regions of animal brains:
YouTube of his talk
Neuroscience has had LOTS of big data; but the data is "information poor":
These results look better than the reconstructions from Thomas’s 2009 paper, but I think the increase in model complexity is much larger than the increase in performance. We’re data limited, people!
I think it will greatly impact AI, in several different ways:
(I would definitely not share something like that with my colleagues -- I would be much more circumspect on what I write, and water it down to the point that it seems uncontroversial. But this is a futurist forum, so see no harm in letting my imagination roam free.)
Even if the most optimistic predictions don't hold up, there are lots of half-steps, quarter-steps, and so on, that will probably pan out.
There is also immense potential to enhance humanity (there are 3 posting in this thread):
Again, there are layers of possible future applications -- some more probable and near-term than others; one shouldn't focus on just a single possible outcome.
That's not the only path to the future development of AI. It happens to be the one I'm most excited to see play out; other paths are discussed at the end here:
I wouldn't want to put a timeline prediction on when any of that will happen -- but it's coming.