I recently completed the book You Look Like a Thing and I Love You: How Artificial Intelligence Works and Why It’s Making the World a Weirder Place by Janelle Shane. I received the book from the Next Big Idea Club – a bookclub that regularly sends a few books that should spark ideas in your life and career (to check it out, go here).
Artificial Intelligence (AI) has been a buzzword for decades actually, which surprises people. If you’re interested in learning more about AI, I wrote a primer for non-technical people to understand this evolving technology here.
Here are a few sections from the book that I found interesting:
- Wikipedia: As we know, humans are biased, and AI trains from data sets that are influenced by humans. Over or underrepresentation of population segments can be exacerbated by algorithms that are skewed. For instance, female scientists are under-represented in Wikipedia compared with male scientists of similar accomplishment. From the book: “Donna Strickland, the 2018 winner of the Nobel Prize in Physics, hadn’t been the subject of a Wikipedia article until after she won – just earlier that year, a draft Wikipedia article about her had been rejected because the editor thought she wasn’t famous enough!”. The idea here is that if an algorithm built to predict future great scientists uses the Wikipedia data to influence how it identifies a pool of high-potential scientists, women would be underrepresented.
- Dolphin Reward-Hacking: AI can be like mammals and figure out how to maximize some reward (or, result) that defeats the intent of the result-seeker. This story is about the mammals rather than AI results, but I enjoyed the story: “Dolphin trainers have learned that it’s handy to get the dolphins to help with keeping their tanks clean. All they have to do is teach the dolphins to fetch trash and bring it to their keepers in exchange for fish. It doesn’t always work well, however. Some dolphins learn that the exchange rate is the same no matter how large the bit of trash is, and they learn to hoard trash instead of returning it, tearing off small pieces to bring to their keepers for a fish apiece.”
- Ignore Your GPS: AI systems lack the sort of context that comes naturally to us. As a result, they suffer from a myopia because they aren’t ready for the unexpected. Consider that during the California wildfires of 2017, AI-driven navigation apps “directed cars toward neighborhoods that were on fire. It wasn’t trying to kill people: it just saw that those neighborhoods has less traffic. Nobody had told it (the AI system) about the fire.”
- Diversity?: On the topic again of AI algorithms training on skewed (biased) datasets, an employment attorney “told of a client who had been screening another company’s recruitment algorithm wanting to discover which feature the algorithm was most strongly correlating with good performance. Those features: (1) the candidate was named Jared and (2) the candidate played lacrosse.”
No matter what, AI will continue to create amazing breakthroughs, generate a lot of controversy both by how it’s used (for instance China developing AI capabilities to control citizens) and how it’s trained (for instance using biased crime datasets to decide how to deploy local law enforcement). It can also become very dangerous as the datasets, deep learning, and sophistication of the underlying algorithms increase. For more reading on how AI can be weird, go to the Janelle’s blog – appropriately named “AI Weirdness”.
As I wrote in the post I referenced earlier, it’s good to think of AI as a “how” more than a “what”. AI is a method for accelerating learning and improving our ability to predict what comes next. Understanding the basics is important as you hear the concept discussed in your industry.