I don't believe we should be as dismissive as AI as he suggests. There is an undertone of populism and a string of weak rebuttals that undermine his argument.

I don't believe we should be as dismissive as AI as he suggests. There is an undertone of populism and a string of weak rebuttals that undermine his argument. Here is why AI is reason for concern: 1. The world is made up of patterns. Proof: mathematics. 2. Patterns can be distilled down to data. 3. Machine Learning is highly effective at analysing patterns. 4. Machine Learning construction itself is a pattern, albeit a very complicated one, by human scales. 5. Learning speed is bottlenecked by hardware and data availability. Both of which are being solved at exponential rates. The other knock is that consciousness and its constituent pieces is a product of evolution, which itself is not outside the realm of physics.(I.e. We're sacks of chemicals) This means it is theoretically possible to reproduce intelligence if the same initial conditions are met. As silicon poses much shorter time gaps than biological substrates, and as equivalent sensory data are fed into simulated environments, the vectors intercepts are quite clear. I think the key is whether we believe machine learning will reach a critical threshold where it will be able to interpret higher-level meaning. There is positive evidence that this has to do with layers of neural circuitry, which recent neural net design has confirmed in output. Additional layers can be added to machines, but it is not so easy to do with humans. The danger is in autonomy and might. Autonomy being higher level cognition mentioned earlier, and might being the ability to bruteforce its way through problems by sheer speed and iteration, which computer excel at. It would be great if you guys can spot flaws in this argument. The conclusions are quite grim at my current outlook. It'll be nice to be proven wrong.