The headlines today (10th October) suggest that a cure for Alzheimer’s is imminent.
Well, not quite.
Scientists at the University of Leicester have arrested brain cell death from prion disease in mice. This is reported to be the first proof that neuro-degeneration can be delayed in any living animal.
Many neurodegenerative diseases involve the production of “mis-folded” proteins, or prions. In Parkinson’s the alpha-synuclein protein goes wrong, in Alzheimer’s it’s the amyloid and tau proteins, and in Huntington’s it’s the Huntington protein. The brain responds by shutting down local protein production for so long that the cells are eventually destroyed. By targeting the way a cell deals with mis-folded proteins, the same drug could cure many diseases.
Unfortunately it will take a long time to translate this achievement from mice to humans, eradicate the side-effects, and then jump through the fantastically expensive and time-consuming regulatory hoops.
The achievement is important, however. And it shows how fast our understanding of the human brain is moving. The ability to scan ever-finer details, along with the ability to process ever-greater amounts of data, is fuelling a revolution in neuroscience.
This has an obvious bearing on AI research. Most AI researchers adopt one of two approaches. The top-down approach is to construct a precise model of the human brain, replicating the mechanism in minute detail, and see if that results in a human-level AI. (Note that this approach doesn’t require you to understand exactly how brain activity creates a human mind: you just need to be able to model it.)
The bottom-up approach is to keep improving and combining existing artificial intelligence systems (e.g. search algorithms, natural language processing, etc). If the systems become comprehensive enough, you end up with an entity possessing human-level cognitive skills – including the ability to convince humans that it is conscious.
The better we understand how the human brain works, the faster both of these approaches will proceed.
