MIT AI Program – What I Learned

Back in October 2018 I started an artificial intelligence program through MIT – a collaboration between their Sloan School of Management and their Computer Science and Artificial Intelligence Lab (CSAIL). The following are some of the high level thoughts and lessons I came away with:

Any AI initiative should align with your organization’s competitive advantage strategy, be it cost leadership, differentiation, or focus.

Core AI technologies are free. Examples are TensorFlow and Stanford Core NLP. So what costs money? Implementation and viable training data. Go to Google and ask them to give you the code for their machine learning framework and they’ll happily oblige (it’s open source). Now ask them for their data and the answer will either be “no”, or “how much are you willing to pay”. The cost to implement AI comes down to technologists and data (if you don’t already have it).

Narrow AI will drastically change our world in the next 10 years. Narrow AI refers to a machine doing a specific task that requires intelligence, like driving a car. Some of the changes that will occur are obvious. Do you have little kids? I bet you won’t ever have to teach them how to drive. Some impacts aren’t so obvious. General AI refers to machines that can actually think and be aware and pursue complex goals. That could be hundreds of years in the future, and as long as humans are able to keep solving problems it will happen. There are no scientific constraints that prevent it. But don’t even bother trying to think of the impact of general AI on society. Just focus on narrow AI because that’s already here. And unlike an infinitely scalable cloud AI network, you only have one brain.

Studies done by MIT have shown that collective intelligence (pairing AI with humans) is the most effective – more than human intelligence or AI alone. What makes a collective intelligence system even more effective? The same studies show that if the human group is diverse, especially in terms of gender, the overall intelligence of the system increases significantly.

Ask a vendor what their AI solution CAN’T do and see what their reaction is. AI is a buzzword and lots of people are trying to capitalize on it.

There is nothing magical about AI. it only requires data, processing power, pattern storage/retrieval, and math (especially vector math).

AI will provide a net benefit to society by creating many new high-tech jobs and eliminating mundane, boring and dangerous jobs.

AI has been around a long time and advancements in processing power, big data, and connected systems are finally making it practical and commonplace. These are exciting times!