Where Can AI Create the Most Value and Impact Today?

On behalf of the Goldhirsh Foundation, Ben Goldhirsh, Claire Hoffman, and I recently hosted a dinner discussion about artificial intelligence with movement shapers and funders in Los Angeles – including heads of large civic institutions, elected officials, journalists, tech leaders, culture practitioners, and investors. (AI, especially as related to ethics and community impact, is a topic we have been tracking at the Goldhirsh Foundation since 2019.)

Our speakers were Jim Gao and Katie Hoffman. Jim and Katie are co-founders of Phaidra. Formerly, Jim worked at DeepMind, generally recognized as one of the leading AI organizations, alongside OpenAI.

While Jim’s and Katie’s expertise is in industrial systems, Jim also discussed AI in protein folding and materials discovery, as well as in breakthroughs related to strategy games chess and Go – all to underscore his points that AI is about creativity, capabilities, knowledge, and improvement.

Our guests were those with the opportunity to address: disability and aging populations, public health, climate change, journalism and misinformation, major transportation networks, and education reform systems. We also hosted a workshop the following day with 50 partner organizations to dive more deeply into how AI can impact work around issues such as criminal justice, environmental sustainability, and economic development.

Jim’s and Katie’s high level thoughts highlighted that there “is a general misconception about where AI will create the most lasting and valuable impact.” Jim noted that the “conversation around AI today tends to center around automation. That’s where most of the fear comes from as well. People tend to see AI within the narrow lens of what existing, routine tasks can AI replace (i.e. automate) today?”

Jim continued: “This is completely wrong. We shouldn’t focus on what AI can do marginally better, or cheaper, than what humans can already do today. That's incremental thinking. We should focus on what new capabilities does AI enable us to do, that we could not do previously? That's where the biggest, most enduring impact of AI will be felt. And I submit to you that the answer is creativity — which I define as new knowledge acquisition. Discovering new knowledge that did not exist before.”

And, similarly: “What they discovered is the power of self-learning systems to navigate immense complexity and fundamentally discover new knowledge that did not exist before.”

Katie’s wisdom to impart is that leveraging AI has the following “key requirements” –

• A clear articulation of the problem to be solved and the ability to define and measure the boundaries in which the AI will operate.
• The objective needs to be measurable — you can't improve what you don't measure.
• Data. AI requires data to learn from.
• Domain expertise. AI cannot be applied safely and in production without deep industry specific domain expertise.

During the conversation that night, the leaders in the room were able to contribute their respective domain expertise around movements and discuss the importance of data being clean and unbiased, policy governing AI clear and ethical, and the utility of AI as a tool in partnership with humans.

These are all values that the Goldhirsh Foundation will continue to keep in mind as we sharpen our focus on AI’s ability to create value and impact now and in the future.