5 things that we are teaching AI since 50s ,60s & other matters
I am watching the video from Andreesen Horowitz on Youtube about the software of tomorrow. The video is a bit dated but the content is still fresh. Below are some of the key points mentioned in the video.
AI needs to be taught the following(These are the use cases discovered for AI back in the 50s and 60s. Only recently do we see their actual application):
- Reasoning in the same capacity as humans
- Knowledge representation for providing interactions with humans
- Planning and navigations (which is already being implemented)
- Natural Language Processing (Initial use case for translating Russians to English and Vice and versa)
- Perception (Teach AI to see things or sense the world — sights, smells and etc.)
- Generalized Intelligence (Emotional intelligence, creativity, moral reasoning and intuition)
The point 6 is important since that’s the last step assuming that the first five points would eventually lead to 6. The primary reasoning is that to reach general intelligence, the machines need to learn the five points that make human intelligence unique in comparison to other species.
As we look at each of the use case that we are teaching the AI, we are seeing better and better AI applications. E.g computer vision in image processing, natural language processing, GPS, AI beating best chess/go players. Man as I am writing right now, it seems by the original vision for teaching the systems is gradually getting closer and closer to actual fruition. I can only imagine how soon we would head to a place where machines would approximate the generalized intelligence — where the Turing test would be passed with flying colors.
This is short note with zero resolution. I ma leave you hanging. Maybe I will come back when I know the answer and then offer peace to this piece. Then again, you’d wait for a couple of years.