One of the initial leaders in Facial Recognition was IBM and, their initial experience (much of it bad), with the technology, helped the company understand the problems with bias. Bias remains one of the significant issues with the development of AI because it can not only make the technology unreliable; it can change it from a tremendous asset to a massive liability.
I’ve been fascinated by the development of AI Facial Recognition because it is one of the key technologies on the path to creating a general-purpose AI. You see, for an AI to fully interact with a human, as a human might, it needs to be able to see and recognize humans just like we see and recognize each other.
I doubt you can do a General Purpose AI without having facial recognition as a feature. This is because so many of the uses for AI will require it to recognize people so it can take what it knows about you to enhance its responses.
So, stepping aside from this technology, for IBM, who remains the leader in enterprise-grade AI, wasn’t a trivial decision.
AI Facial Recognition
This technology didn’t start well, and it quickly became the poster child for how bias could destroy the effectiveness of artificial intelligence.
IBM had a massive effort to fix the technology, a technology critical to the eventual future of their AI program, and they still just walked away from it. That is significant, and the reason is that they became aware that the technology was being abused.
This move would not have been a trivial decision. Still, IBM husbands their brand aggressively, they have been a leader in driving diversity into the industry, and they have one of the most diverse employees and executive teams currently in the technology market.
Given the investment and progress with the technology, to get them to abandon, it required a clear and present danger that it would be used in a way inconsistent with IBM’s brand, ethics, and diversity efforts. This potentially devastating move suggests that facial recognition technology is being used broadly against minorities, maybe intentionally biased, and that the exposure was possibly so extreme that it overcame IBM’s need to continue to develop it. A need tightly tied to IBM’s long term future, and IBM is also one for the most strategic companies in the segment.
While the benefits of the technology remain, much like any powerful tool, there needs to be a strong effort to regulate and assure it isn’t misused. Without that regulation, companies with a strong moral core will have to abandon it until such time they can be assured it won’t be used in a way that will damage the firm’s ethics or long term Social Responsibility goals.
IBM was given a hard choice between its long term business goals and doing what was right; most companies tend to make these choices poorly; IBM did not. They stepped up at great sacrifice, we can only hope they set a strong enough example, so others do as well, forcing governments to regulate this potentially amazing, but currently abused, technology.