My Call to Responsible AI Action

OK, I’m doing it. I’m putting pen to paper, or perhaps better put, pixel to screen.

Eight months ago, I moved from Boulder, Colorado to San Francisco. I’d just finished my master’s degree in Information Science, which I like to describe as data science with a conscience. I knew how to code, and I understood the massive potential for conventional personal success involved with getting into the AI industry. I also recognized the cavalier bravado with which the Altmans and Musks of the space were pushing AI development and deployment. Their statements suggested an investment in public benefit — AI would usher in a new utopia in which every global problem was resolved with unquestionable nouse. Their actions, however, suggested to me a desire for capitalistic gain and an accrual of unprecedented power. The true intentions of these actors were unclear, but my personal intentions became obvious. I entered the epicenter of AI with one broad goal in mind: to hold the largest and most powerful developers accountable for the cataclysmic effects of their pursuits.

In the months since, numerous interactions have continued to shape my perspective on what’s happening in this city. Many of my Gen Z peers employed at AI startups based here have delightedly shared their stories of contributing to their companies’ growth and prosperity. I fostered connections with individuals working in regulatory compliance at some of the largest AI firms, who politely let me know that my ethical AI aspirations might render me an “unhireable” candidate. My conversation over breakfast with a venture capitalist heavily invested in the space reminded me of the unavoidable bottom line that often drives innovation. Of all those that I’ve bumped elbows with here, one individual I met three weeks ago certainly holds the most influence over the trajectory of AI development. That would be Dario Amodei, the co-founder and CEO of Anthropic.

My exploration of these issues, while fascinating and fulfilling, has yet to afford me the financial security necessary for paying the rent demanded of a San Francisco lease. As such, I work four days a week as a server at a renowned restaurant near downtown. On a rainy Friday lunch shift, Mr. Amodei found himself in my section at a quiet outdoor table. His arrival was a surprise to the staff, as the reservation had been made for John Thornhill, the innovation editor of the Financial Times. It turns out the two had chosen to conduct an interview over a meal. As a product of the drizzly conditions, my half-empty section allowed me to eavesdrop on a majority of their conversation. The article produced from this interview has now been published, which you can find here.

Anthropic is well known for its recent pushback against the US Department of Defense, with the company putting its proverbial foot down regarding use of their technology to develop autonomous lethal weapons or conduct mass surveillance. Amodei seemed keen to reinforce his position as an ethical AI representative during the interview, differing himself and his company from other leaders in the industry. That said, I found a few of his statements throughout the interview to be in conflict with that position. Or, more explicitly, his statements demonstrated a prioritization of his company’s standing in the brawl for AI dominance, with altruistic intentions seemingly leveraged as PR devices for public support.

One of my major concerns regarding AI is the breakneck speed at which it is being developed and deployed. Considerations of the adverse effects of new product rollouts appear to be at best an afterthought. When asked about whether development was reaching a plateau, Amodei responded that “there’s no end to the rainbow. There’s just the rainbow. We don’t see anything slowing down.” Does that statement suggest Amodei and Anthropic’s commitment to responsible, meticulous development? I posit that it adheres far more faithfully to industry norms of speed over precision, of opportunism over scrupulous decision-making.

The way in which I originally interpreted Anthropic’s DoD conflict itself was also largely undermined by Amodei’s words. Amodei claimed that he’s “excited for the US government to use this technology,” suggesting that it could be wielded as a tool for dismantling autocracies and authoritarian regimes abroad in the name of protecting democracy. How that dismantling might take place was unclear. Regardless, doubt was cast over whether Amodei truly cares about ethical governmental uses of AI or merely his company’s association with unethical behavior.

Amodei’s desire to propagate democracy is noble. Nevertheless, when I consider democracy as it relates to AI, my primary feeling is one of fear. Did the people elect a handful of billionaires — including Amodei — to make critical decisions on how AI is integrated into society? The free market did, perhaps. That’s a valid answer. My concern is that the “broligarchy” that Amodei belongs to (despite his attempts to differentiate himself) holds immense influence over the future of the human race without ever having been formally selected for positions of such power. The scale of that power appears to increase day by day. Tristan Harris outlined our direction of travel towards a growing power imbalance in his recent podcast interview with David Fenton. As humans are replaced in the workforce by AI, the strength of labor unions is reduced, as are the tax dollars produced for the government. The bargaining power of the public is under threat, and I’d argue that the foundations of our democracy are inextricably linked to that phenomenon.

To you, reader, I encourage you to investigate the alignment of statements and actions by powerful AI actors like Dario Amodei. Are they truly demonstrating the ethics they claim to support in their actions? Is our trust in these stakeholders being earned honestly, or being deftly manipulated? I don’t purport to have the answers, but I’m certain we need collective vigilance to find them.

Thank you for reading my first article. More to come soon.