A Perry World House Student Fellows Policy Project
What is AI?
That's how Andrew Ng, a leading expert on AI, might answer the question. In fact, this answer has been published ad nauseam around the world. But what technology underpins AI? Why has it advanced to the point that Prof. Ng can make such a sweeping claim? This section will address some of the underlying trends that have enabled such rapid progress in AI over the last 5 to 7 years.
A talk by Prof. Ng where he discusses his proposition in detail.
Let's start with the question of what AI really is. Unsurprisingly, there are a lot of subtleties to this question. Defining intelligence itself can be challenging; how can we define artificial intelligence?
Andressen Horowitz AI Talk
Frank Chen, a partner at A16Z, has been a prolific speaker on the topic of what artificial intelligence is and how it will change the world. This 45 minute talk is an excellent discussion on what AI is.
AIMA: A Modern Approach
This table is pulled from the most famous textbook on artificial intelligence, Russell and Norvig's AIMA. It highlights all the different kinds of definitions for AI that have proliferated over time
In our view, the best definition is Russell and Norvig's classic assertion:
There are a few important and positive things to note about this definition:
1) It focuses on intelligent agents, an expansive definition of what can be achieved in AI.
2) It does not subject AI to any specific set of tests, preferring instead to use the perception-action framework.
3) It is technology agnostic and capable of encompassing the vast number of subfields in AI.
Speaking of subfields... AI has a lot of them!
Will this site provide information on how each of these fields is progressing? No!
While it is helpful to see all the different fields that contribute to progress in AI, the reality is that tracking progress in each of these fields is a monumental task, let alone across all of them. If this is still of interest, here is a gallery that provides some excellent summary resources on the progress of each field.
NLP (Natural Language Processing
At the end of the day, there are a few key trends that are enabling the progress across all fields of AI.
We highlight two trends we believe are forming the technical basis for the advances in AI: deep learning in software and revolutionary chip technology in hardware.