Replies

(Note here that we’re talking about “App AI” such as ChatGPT or the art AIs, which are limited to a single area of activity. General Intelligence AI, the HAL 9000s and Skynets of the movies, doesn’t exist and likely never will.)

Any author that thinks that AI will plateau at the level of ChatGPT doesn't understand AI.

Whether or not it ever reaches sentience is another matter altogether and one subject to pure speculation. But that aside, AI is progressing at light-speed.

For example, as crazy as it sounds, Stephen L. Thaler, PhD, president and CEO of Imagination Engines, thinks he has already created an LLM that is sentient.

‘It has feelings’: Inventor says he has built a sentient AI

Thaler may be blowing smoke, but the point is that scientists are going to be pushing the edge of the envelope re: AGI. We may never get there, but we will keep getting closer and closer.

BTW, ChatGPT already looks primitive compared to some of the other systems coming on board. When it comes to controlling AI, another problem arises. Increasing the power of AI means increasing the size of the data sets it consumes. Control means smaller data sets (i.e., the elimination of unwanted data). A controlled system is, by its very nature, a less accurate, less efficient, and less powerful system.

For example, there have already been complaints that AI will lead to racial discrimination by banks re: lending to minorities. Banks can't make AI systems ignore people with poor credit despite their minority status. In other words, AI won't discriminate in favor of minorities unless the banks feed it tiny biased data sets (favoring minorities by ignoring their poor credit)...and tiny data sets ruin the AI systems' ability to operate efficiently.