Conversation

If you insert electric probes into an insect *before* adulthood, its tissues can organically grow around the probe and unlock a high-bandwidth insect-machine interface. Then you can read data from the insect's brain and *control* its flight by stimulation. This is from 2009, but I believe modern hardware and AI stack can dramatically improve the "insect cyborg": - Microchips and onboard computing are now way more powerful. - Diffusion is able to reconstruct images from even the human brain, let alone an insect's. We will be able to decode the insect's vision and plan accordingly. - Control methods are more sample efficient & robust. With enough insect samples, we can gather data to train a highly nontrivial flight controller. RLIF = reinforcement learning from insect feedback?
Image
Image
Note: I'm just pointing out the technical feasibility of this. People thought about it in 2009, so it isn't a new idea. The technology is well within reach - IMHO < 5 years if someone works on this seriously. Now whether we *should* do it is a completely different issue.
5
85
's Neuralink is of course a MUCH more scaled up version of this: unlocking the brain-machine interface for humans. I don't think human-level is within reach yet, but we will solve insect's brain first. Again, massive opportunities come along with massive risks and… Show more
5
74
Interesting (and scary). Post maturation brain is too complicated for us to hack into. But put the interface during growth (in-embryo?) and the brain integrates itself with the interface. As incredibly complex and intricate nature is, humans are just as ingenious and relentless.
2
22
Show replies