Tech executives responsible for AI ethics – but not personal actions?
• AI ethics commitments made by major players.
• Notably, attendees at the White House were all male and mostly white.
• There’s a potential for gaslighting double standards in the people deciding AI ethics.
AI ethics have been under debate since before the general public was familiar enough with the technology to understand the abbreviation.
Last Friday, a public commitment to promoting safety and transparency in the development of artificial intelligence was made by US technology companies including Google and OpenAI.
Coming together at the White House, executives from Amazon, Anthropic, Inflection AI, Meta, and Microsoft made “voluntary” commitments to “help move toward safe, secure and transparent development of AI technology.”
According to Anna Makanjju, vice-president of global affairs at OpenAI, the commitment will “contribute specific and concrete practices” to what she called an “ongoing discussion” about AI regulation.
As well as committing to more security testing, the companies also pledged to share more information across the industry and government about how they’re mitigating risk.
The Biden administration and Capitol Hill lawmakers – like governments globally – have had to scramble together a coherent policy response to the rapid emergence of AI technologies.
The White House has called the commitments a “critical step toward developing responsible AI,” noting that an executive order was still in preparation, and urging Congress to pass legislation.
Who was (and wasn’t) there
Among the executives who appeared at the White House alongside president Biden to tout their new public undertakings were Microsoft president Brad Smith and Nick Clegg, the president of Facebook and Instagram parent Meta.
Many noticed the entirely male lineup at the White House last Friday. This is somewhat ironic, given the number of women leading the fight for a more ethical approach to AI adoption.
Earlier this year, UNESCO launched a Women 4 Ethical AI Platform, stating that “without an ethical framework, the development and deployment of AI threatens to reproduce gender inequalities present in the real world, and even magnify them.”
It’s not as though there aren’t women at the top of AI companies, either. The turnout at the global summit of AI for Good, held in Geneva this year, is proof of that: the speakers included Joanna Shields (Baroness Shields OBE), Chief Executive Officer of BenevolentAI.
BenevolentAI is the world leader in the development and application of artificial intelligence and machine learning to understand the underlying causes of disease, accelerate drug discovery and develop new medicines for hard-to-treat diseases.
Also arguably overlooked were the women working in the American government on the very legislation that the White House is urging Congress to provide. Anne Neuberger, an American national security official who serves as national security advisor for cyber and emerging technology in the Biden administration, was present at the summit – but not the White House.
Google DeepMind’s COO is Lila Ibrahim, who has previously worked at Intel. Given her responsibilities at the company, which include Governance and Ethics, she would have been an ideal attendee last week.
Instead, Google’s president of global affairs was present at the White House on July 21. Also there, and likely a familiar face for Kent Walker, was Mustafa Suleyman.
Last Friday, when Suleyman was at the White House he was representing Inflection AI, where he is currently chief executive. He’s prominent in the discourse surrounding AI ethics and speaks widely about the need for technologists to be held responsible for their work.
What’s been conveniently forgotten is his reputation at Google, where he faced little responsibility for actions of his own.
Suleyman co-founded DeepMind, which was acquired by Google in 2014. His management style was the source of complaints from DeepMind staff who, Insider reported, had filed grievances over several years.
“He had a habit of flying out of nowhere,” said a former employee. “It felt like he wanted to humiliate you; like ‘I’m trying to catch you off guard,’ he would just start messing with you, in front of your colleagues, without any warning. ”
People familiar with the matter believed that Suleyman was aware of the effect this behavior had on employees.
“He used to say, ‘I crush people,'” said a former employee.
Apparently, there were confidentiality agreements made between DeepMind and former and current colleagues who complained about his management.
In August 2019, Suleyman was placed on administrative leave – at the time, a spokesperson said Suleyman was “taking a break after 10 busy years” – and an external lawyer was hired to investigate bullying allegations. Shortly afterwards, he left the company.
Ah, justice? Nope!
Mustafa Suleyman was offered, and took on, a VP role at Google, promoting him to a position just two steps below the CEO. In early 2022, he left Google for Greylock Partners, a venture capital firm that saw him through to attending the White House on behalf of Inflection.
When he left Google, on being asked about the bullying mess, Suleyman’s response was an apology: “I really screwed up,” he said. “I remain very sorry about the impact that that caused people and the hurt that people felt there.”
AI regulation and diversity issues
Both staff members had been calling for more diversity among Google’s research staff and expressed concern that the company was censoring research critical of its products.
“Not only does it make me deeply question the commitment to ethics and diversity inside the company,” Scott Niekum, an assistant professor at the University of Texas at Austin who works on robotics and machine learning, told The Verge.
“But it worries me that they’ve shown a willingness to suppress science that doesn’t align with their business interests.”
Gebru claimed she was fired after questioning an order not to publish a paper claiming that, to paraphrase, AI threatens to deepen the dominance of a way of thinking that is white, male, comparatively affluent and focused on the US and Europe.
Ironically, the language Google used to announce her departure – “behavior that is inconsistent with the expectations of a Google manager” – echoes that used upon Suleyman’s withdrawal from DeepMind.
The email sent to staff in 2019, after the news of Suleyman’s departure broke, read that his “management style fell short” of expected standards.
Now, this might be old news: the issue hit headlines in early 2021. However, have more recent headlines not vindicated the case she was making? Not only has the intrinsic bias of artificially generated text been proven, but so has the exploitation of already marginalized groups that came when AI companies (reactively) tried to stop their chatbots’ racism.
AI ethics: coming full circle
It might be that, at first glance, press photos from the AI ethics meeting at the White House showed nothing unusual. Perhaps the people most involved in AI ethics are all men! Famously, groups of (predominantly white) men have been at the forefront of ethical change and equality…
After years of silencing anyone who pointed out the human dangers of AI, technology entrepreneurs have got on board with criticizing the potential harms of AI. However, the approach only gives a platform to “AI Will Destroy Humanity – Robot Domination Imminent” headlines.
The lack of diversity at the White House meeting is compounded by Suleyman’s attendance. The people deciding what AI ethics is concerned with are themselves ethical nightmares. Convenient focus on AI’s world takeover means Google has been able to sweep allegations of misconduct under the carpet.
A Google search for Mustafa Suleyman, even with the keyword “bullying” does little to expose his history of manipulation and humiliation.
While those in power pull together to stress the importance of responsible AI adoption, they are also pulling the curtain over their own unethical practices and aspirations.