Artificial Intelligence Editorial

The bias in AI *influencers*

One of the challenges that we’ll face in AI is bias — not just in the data, but the influencers as well. Consider this tweet from @OnPageLeads:

@OnPageLeads’ tweet.
Tap to view the original tweet.

Take a closer look at the original drawing made by the child…

The original child’s drawing of the hand. Make note of the skin tone.
Tap to view the source.

…and then the AI-generated photorealistic image:

The AI’s photorealistic rendering of the child’s drawing.
Again, make note of the skin tone.
Tap to view the source.

Tech influencer Robert Scoble, societally-blind gadget fanboy that he is, was quick to heap praise on the tweet. Thankfully, he was quickly called out by people who saw what the problem was:

Robert Scoble’s poorly-considered response, followed by some righteous retorts.
Tap to view the original tweet.

Of course, this kind of structural racism is nothing new to us folks of color. The problem is criticism of this kind often gets shut down. The most egregious case of this was Google’s firing of AI ethicist Timnit Gebru, who has warned time and again that unmoderated AI has the power to enhance societal racism.

You may have heard of recent ex-Googler Geoffrey Hinton, who’s making headlines about sounding the alarm about possible existential threats about AI. He was oddly silent when Google was firing Gebru and others for saying that AI could harm marginalized people.

In fact, he downplayed their concerns in this CNN interview:

“Their concerns aren’t as existentially serious as the idea of these things getting more intelligent than us and taking over.”

Geoffrey Hinton, CNN, May 2, 2023.

My only reply to Hinton’s remark is this:

The Audacity of the Caucasity

For more about Timnit Gebru and her concerns about AI — especially a lack of focus on ethics related to it — check out this podcast episode of Adam Conover’s Factually, featuring Gebru and computational linguistics professor Emily Bender: