Earlier this year, I took a course on artificial intelligence taught at the Executive Education division of the UC Berkeley Haas School of Management. The course was primarily focused on AI business strategies and applications.
The course covered many topics, including:
- The basics of machine learning
- Neural networks
- Deep learning
- Computer vision
- Natural language processing
- AI strategy
- Building an AI team
- The future of AI in business
It was a fascinating course taught by a collection of AI professionals. Some were on the technical side of AI development, some on the business side, and one focused primarily on ethics. I wish I could say there was one big “Aha!” moment for me. Instead, the course was a combination of new information and a confirmation of the many things I’ve written and spoken about in the AI arena.
I’m intrigued by AI not just because it is inherently intriguing, but because I see its impact on our industry. You’ve probably read my AI chapter in The Personalization Paradox. Don’t worry. I don’t expect you to have it memorized. (Isn’t that what computers are for?).
In the book, I share how AI engines work and how they are already in use today. I describe how standardized, structured content is crucial to training AI engines. An AI engine can be key to delivering personalized experiences for customers on a massive scale.
The AI is always learning and it can adjust the way it acts on information over time. The customer experience gets better and better, and your company can sit back and reap the profits. But only if your content is ready to be retrieved, assembled, and delivered in such a way.
I’d say that the most surprising thing I learned from the course is just how much AI technology has advanced over the past 5-10 years. The study of artificial intelligence goes back decades. But in the past 10 years, advances have happened fast and furiously. The reason for the acceleration of advancement is two-fold:
- Big data – AI systems need to be trained with data (content). Now that we have enormous quantities of content, we can train the AI engines with enough information for them to be useful.
- Processing power – AI systems require a tremendous amount of processing power. Advances in chip technology have provided the raw computing power that AI engines need to do their work.
Some people are predicting a slow down in the speed at which advances have been made. I’m not so sure I agree. There are many efforts underway to increase compute power either by shrinking transistors on chips more or by using a “network” of chips to do the work.
I think my favorite line from the course is, “Artificial Intelligence is just complicated math.” Which is true, of course. But my oh my, the math is very complicated, indeed!
This article was adapted from Val’s CEO Message in the Content Rules email newsletter. For more exclusive content directly from Val, subscribe below! We send about two issues a month and we never sell, rent, share, or otherwise show your email address to anyone.