AI Ethics & Policy Insights

The AI ethics & policy info we should know - simplified. Week of November 27 , 2023. Issue 17.

Paige

Nov 30, 2023

I took last week off for the American Thanksgiving Holiday, but I’m happy to be back!

Major news this week:

OPEN AI HAPPENINGS
I’m certain most people know about the Open AI chaos, but if you don’t, here’s a recap.
- The Firing: On Friday, November 17th, the six-person board of Open AI opted to fire Sam Altman stating that he had not been “consistently candid in his communications with the board”.
- Turning Tables: Sam appeared to be truly blindsided, stating on X that he would have updates on great things coming soon. Greg Brockman, a member of the board and co-founder of Open AI was not party to the vote, leaving Sam’s fate in the hands of four people, including Ilia Sutskever, fellow Open AI founder, who voted to oust Sam.
- Solidarity: Greg Brockman was originally staying on with Open AI but stepping down from the board. Soon after this announcement he opted to quit in solidarity with Sam. Greg also posted on X that something great would be coming soon.
- The Peace Broker: Satya Nadella was shocked by this news and spent his pre-Thanksgiving weekend trying to broker a deal to get Sam back as CEO of Open AI. The deal came close several times and fell apart, so Satya offered jobs to Sam and Greg to lead a generative AI lab at Microsoft. Some people also stated that Microsoft offered jobs to the 700 Open AI employees who signed a letter saying they would quit Open AI because of the board’s handling of the situation and their loyalty to Sam. This made some of the 10,000+ laid off MSFT employees pretty upset.
- Turning Tables…Again: Ilia Sutskever, one of the founding members of Open AI who voted to oust Sam Altman, stated that he regretted how things happened and decided to sign the letter with 700 signatures threatening to leave Open AI if Sam wasn’t reinstated. I think we’re all wondering - what changed between his vote and his signature?
- The New CEO: The board had said they were beginning the hunt for a new CEO, and then three days later announced that the former CEO of Twitch would be the CEO. He posted about it on X and announced that he would be taking the role.
- The Reinstating: Sam was officially reinstated by Tuesday, November 21, and the creation of a new board is under way. You can read Sam’s letter to the team after his reinstatement here.

Women in AI - are they hard to find?

Spoiler: the answer is no, they’re not hard to find.

The Context:

In recent weeks I’ve seen an uptick in posts and controversies related to women in AI. Some people are saying that they can’t find women in AI, and many are saying that if you can’t find women in AI, you’re simply not looking. This has all come to a climax as major names pulled out of a developer conference when it was found out that the female speakers listed on the website weren’t real.

Conference Controversy:

The DevTernity conference is over, and execs from both Microsoft and Amazon have dropped out after it was reported that at least one of the female speakers listed on the conference’s speaker lineup was a fake woman, listed as an employee of a real company. The conference organizer Eduards Sizovs stated that the profile in question was an “auto-generated” female profile created after another speaker dropped out, and is refusing to apologize for the blunder. Instead he blamed someone for trying to “deliberately damage the conference”. But this appears to be a pattern. Research by Gergely Orosz shows that the DevTernity conference “lists fake (invented) Coinbase & Meta engineers as speakers on four conferences in 2021, 2022, and 2023.” This was all made worse by the fact that Sizovs does not ever call women “women” but rather refers to them as “ladies.”

The Conversation:

This is an important conversation to have, and while progress on the “women in tech” front is slower than I’d like, I’m grateful that people with platforms and voices are taking the time to speak up.

In the wake of two female Open AI board members being replaced by men, an article in The Guardian boldly asked the question “Where are all the ‘godmothers’ of AI?” This is an important question, as the ‘godfathers’ of AI are regularly cited in articles and papers the world over. But nobody has taken the time to officially bestow those titles on the women who have helped make AI what it is today.

In response and support for this article David Polgar, founder of All Tech Is Human, wrote a blog post on his organization’s website entitled “If you’re not utilizing women leaders in AI, you’re just not looking hard enough.” The point of his article cannot be missed or mistaken; Polger states that it’s not whether there are women in AI - it’s whether you’re listening to them. He goes on to list a number of well-known, accomplished females who have shaped the AI landscape, and points to the 100 Brilliant Women in AI Ethics list as resources to find female voices in AI.

Professor Casey Fiesler, an amazing content creator at the intersection of tech, ethics, law, internet and academia, posted a video reviewing the conference controversy discussed above also did a deep dive on the “coding unicorn” in this video and asked some critical questions about what it means for women in tech if the account isn’t really created by a woman.

Why it matters:

We have to keep having the conversation about women in AI and women in tech. As of now, only 25% of the people working in AI are women. Having female voices in every stage of AI - from the idea to the deployment and governance - is incredibly important. Outside of gender, there are a number of voices from all different cultures, perspectives and worldviews that are contributing significantly to the world of AI.

When someone says to me “there are fewer women in tech and AI, so it’s harder to find them” I have a hard time believing that claim. I think these women are easy to find - because we’re out here creating content, writing books, sharing thought leadership and more - but the standard by which men and women are judged and therefore selected for opportunities is different. A 2014 Harvard Business Review study found that men are often judged or selected for opportunities based on potential and what they could bring to the table, while women are judged and selected for those same opportunities based on proof and what they’ve already done.

The ethical implications:

Without the incorporation of female voices in AI development, ideation, policy, creation, research and deployment, we risk creating less fair, less transparent, more dangerous technology that will cause more inequity and harm than technology of the past. Incorporating women in AI into all aspects of the AI product lifecycle, rooms where AI is being discussed, and where AI regulation and policy decisions are being made isn’t a friendly D&I opportunity, a box to check or a favor you’re doing for women - it is critical to the safety of humanity. If there are no women in your working groups, it’s not because we’re not out here. It’s because you’re not doing the work to bring them into the conversation.

Sources: The Guardian, All Tech is Human, Business Insider, Futurism, AP News, Harvard Business Review

AI Policy Updates

The US Department of Defense Endorses RAI for Global Militaries

In February of 2023 the US government launched the “Political Declaration on Responsible Military use of Artificial Intelligence and Autonomy”. The State Department announced that since that launch, 47 states have endorsed the declaration, which sets out non-legally binding guidelines and best practices for AI use by a military. Read more here.

Meta has Updated its Political Ad Rules on the heels of Google

Following the example set by Google, Meta has updated its political ad rules for the AI era, and now requires political advertisers to disclose if they’re using AI to “to create or alter a political or social issue ad in certain cases.” Meta faces a significant challenge in the run-up to the next major US election as it consistently faces backlash for the role that Facebook played by allowing the significant spread of mis and disinformation on its site during previous elections. This policy doesn’t ban the use of generative AI technology, and it’s unclear how prominent the disclosure must be, or if the advertiser gets to determine what the disclosure is. Read more here.