AI Is Nothing Like What Is On The News And Here Is Why

Summary of the Article

In an article written by Andrew Ng(article found here), he actually explains what AI can do right now and how it is nothing close to what we are all seeing in the news. Ng’s main point is that AI can only do simple tasks right now and even those tasks take thousands of hours of examples for the AI to do said task consistently and correctly. Ng explains that right now AI can only do A–>B tasks. This means that AI can only take some input and reference it with its training and then return some output B (the technical term for this is supervised learning). Ng further points out that even this simple form of AI takes tens of thousands of hours of audio or hundreds of thousands of pictures to train. As far as the application in the business world the actual problem isn’t the code – since AI code is pretty open source nowadays – but rather the talent. For AI application into business the business needs to customize the code and train the AI specifically to its business doing this is very time-consuming, very costly, but most importantly requires a level of human capital that only a very small number of people have and it’s that talent that is in so high demand since just copying some code and running some data through it won’t work for AI.

My Take

This article was reliving due to the amount of negative news I have been seeing recently about AI. However, on the other hand, it was extremely fascinating and thought-provoking. If most of the AI applications to date are just from A–>B processing imagine the possibilities when we move on to higher-level types of AI and it becomes readily available. The possibilities are endless! It was also intriguing to me to learn about supervised learning and what that actually means as well as its applications. The author has a number of different accolades in this field and you could tell just by the way that he presented the information, it was clear, concise, and understandable.

Conclusion

All in all, this was an extremely reliving, and fascinating article. It introduced me to some of the technical sides of AI and the context of said technical content at the same time. The author did a really good job explaining the concepts in an easy-to-understand manner. Overall it was a really good article and I recommend you check it out (the article can be found here).