Summary of Talk
In a TED talk given by Andrew Ng(can be found here), he talks about how AI in its current moment is only being used for the big “one size fits all” projects and there are a number of smaller projects that AI could be used to help. The reason that this is the way it is simple – is money. Right now, big tech companies have to pay really expensive engineers to help design AI systems and the only way for that to be economical for them is to have them work on an AI project that can be used for millions of people in one go. For example, social media ad tailoring is one algorithm that controls what ads you see on Instagram for everyone. Compare making that to making an algorithm for a local pizza shop on which flavors to make, or a local hospital on when to order what medical supplies, on the surface the local projects might be cheaper but that fails to consider that the social media algorithm is used for hundreds of millions of people per day while the local projects get maybe thousands. When you consider this fact, it is clear to see which way the economics tilt in this day and age. But that’s just what Ng argues – in this day and age. Later when AI becomes cheaper to produce that is when the combined value of all of these smaller AI projects will be acted upon.
My Take
This made a lot of sense to me and was very interesting to me. It answered the question “if AI is so useful why don’t we use it for everything?” – the answer as always – is money. The relation to data science is evident, it is: the sheer amount of data simply involved in AI, Ng said it himself – AI is really good at recognizing patterns if given the right data. I really liked this talk and how it was put in such an easy-to-understand way. Ng took the example of a pizza and T-shirt shop, both of which are easily imaginable, relatable, and understandable, for almost any given watcher.
This talk also really showed me why people are calling data the fifth industrial revolution. Its applicability just to everyone’s life and profitability are unmatched by anything other than the likes of the other 4 industrial revolutions. When as many people as possible are fluent in this, data will be such an integral part of our lives it will be, in my opinion, just as necessary as electricity. The reason for this is it can help us see the future with the tools of today, make informed evidence-based decisions, and going forward that is going to be what is going to propel the world.
Conclusion
All in all, until now we haven’t been able to tap into the true value of data which lies in the smaller projects We haven’t because companies are focused on doing the one size fits all projects and when the smaller projects become profitable for them in the future they will go onto those which is where the true value of all of this technology lies. Not only will we help the economy tremendously with those projects but we will also help the person who owns the local pizza shop or local hospital to be more profitable and help give back to the community. All of this and more is why data is being dubbed the next industrial revolution. I hope you are just as excited as me for all of this to come. Check out the ted talk here.