AI is not going to kill you!
Everyone thinks Skynet is imminent. Elon Musk is making sensational headlines as always. This annoys me because it ignores something fundamental about modern AI and AI advancement: modern AI is not Artificial General Intelligence.
There is a glaring difference between the two.
AGI is the holy grail of artificial intelligence. It is the end-goal. Arguably, we have made little to no progress in realizing AGI in the past few decades. The recent victories with deep learning is an example of applied AI, or narrow AI. This approach attempts to solve very specific problems. Recent advances in narrow AI include:
- speech recognition
- handwriting recognition
- object recognition
- machine translation
- machine captioning
- natural language processing
- AlphaGo
Yes, deep learning is a recurring theme in modern AI, but that does not mean that deep learning is AGI or even remotely close.
AlphaGo cannot go out and get itself a job or a date. All neural networks thus far only produce useful and meaningful outputs for the inputs they were trained for. Not only that, but the architecture of the network and the hyperparameters are significantly different from problem to problem.
AGI at this point seems like a hopeless dream. Far from being able to simulate a human brain, I think we are struggling to simulate a worm with less than a thousand cells total.
https://en.wikipedia.org/wiki/Artificial_general_intelligence#Current_mainstream_AI_research http://www.openworm.org/