Meanwhile, this DataRobot study takes a darker view, reporting that 54% of tech leaders are concerned about AI bias, up from 42% in 2019. A surprising (to me) 81% think government regulation would be helpful even though 45% also worry regulations would make adoption more difficult. In a classic “not my problem” result, 90% are at least somewhat moderately confident that their own organization can identify AI bias.
Do you miss those innocent days when every product announcement breathlessly reported the new system or feature was “AI-based”? If so, here’s a whiff of remembrance from AI infrastructure provider SambaNova Systems, who found that 78% of large companies rate AI and Machine Learning as important revenue drivers. Yet even they have a curiously mixed attitude, citing (in their press release) a NewVantage Partners study that found only a quarter of top companies have scaled AI/ML across their organization.
Still dark and closer to home for marketers, IBM looked at AI bias in programmatic advertising. Working with industry groups, they’ve released an open source toolkit AI Fairness 360 that can detect it. They concluded potentially harmful bias does exist but were optimistic that technical cures could be found, although ad efficiency might suffer.
It’s tough to get brand safety right, but the industry keeps trying. IPG Mediabrands announced a new set of tools to find and block inappropriate ad placements, while IAS expanded its suitability measurements to include standards from the Global Alliance for Responsible Media (GARM). Meanwhile, DoubleVerify admitted a mistake made brand safety on X/Twitter look worse than it really was in October 2023 and March 2024.