AI on the Horizon: Challenges, Directions, Futures

Eric Horvitz is an absolute authority on technology. For the last 25 years, he’s worked at Microsoft. Now, he currently serves as the technical fellow and managing director of Microsoft Research.

Horvitz said that artificial intelligence is at a tipping point, where it will begin to become a more common staple in our lives, as both computer hardware and software learning and reasoning have improved significantly.

Initially, Horvitz mainly focused on potential benefits that AI could bring to the world. His goals are to create AI that will “assist, empower, and protect.”

Horvitz said that the third most common cause of death in the United States is hospital deaths and preventable errors.

In a not too distant future, AI could assist in reducing these types of deaths. With AI’s reasoning, it could help doctors predict which patients are likely to be readmitted to the hospital. In other cases, AI could control robotic arms that assist and double check human doctors.

In the case of self-driving cars, Horvitz discussed the ethical implications. AI would be expected to make split-second decisions that could put people’s lives at risk.

Horvitz said that self-driving cars could potentially put a significant amount of Americans out of work, such as truck drivers, taxi drivers and delivery drivers.

An example of AI being manipulated to identify street signs incorrectly

Bad actors can potentially manipulate AI so that it behaves incorrectly. There was already such a case with a Twitter bot Microsoft created. In this case, bad actors were able to manipulate the bot into going on racist tangents.

In other cases, AI has shown an embedded racial bias.

Horvitz said that AI often ranks African-Americans as more likely to be criminals than other races.

Thanks again to Eric Horvitz for being part of the SXTXState project. Don’t forget to follow him on Twitter.