Stephen Brobst is a man on a mission. The Chief Technology Officer for Teradata Corporation, he also served on Barack Obama’s Presidential Council of Advisors on Science and Technology (PCAST), helping with the development of the Big Data strategy for the US government. He was a keynote speaker at the European Association of Distance Learning (EADL) conference in Copenhagen earlier this year. The opening salvo of his presentation was that deep learning is not AI Magic, it is AI Math. And so, we were off and running….
AI in Education
He pointed out that AI personalised tutors have had most success in maths and science subjects where the answers are right or wrong. However, they still have a long way to go in any of the humanities, arts and social sciences areas.
Bias in Datasets
An enthusiast for Artificial Intelligence, as we are at Mettacomms, cautioned about how bias, and the perception that AI might be biased, will cause approx. 10% of people to turn away from technology in the future.
He pointed out that bias can be present, even when you do your best to eliminate it. The choice of datasets to include, the size of the data sample or simply the (human) decision about what to include or exclude are all areas of bias in AI. He also pointed out that bias can be the result of time or money pressures.
AI will not replace Humans.
Stephen pointed out that AI doesn’t replace people, it replaces skills. He adapted a quotation from the Harvard Business Review, which I will paraphrase if you will forgive me. Basically, their point was that teachers will not be displaced by AI, rather, those teachers using AI (who are, presumably, much better informed) will replace those who do not use it.
Our CTO, Dr. Emer Gilmartin, has been conscious of bias since day one when we started building Mettacomms. We should get these two together at some point. They have a lot of expertise in common and we could all learn from their insights.
