Main content

Team 24 - AI pioneer Dr Fei-Fei Li speaks up for humanity

Phil Wainewright Profile picture for user pwainewright May 2, 2024
Summary:
AI luminary Dr Fei-Fei Li is a crucial contributor to the tech innovation that ultimately led to generative AI. Her take on its impact emphasizes our humanity.

Dr Fei-Fei Li seated on stage between co-panelists speaks at Team 24
Dr Fei-Fei Li (center) speaks at Team 24 (@philww)

With AI dominating the agenda at Team 24 this year — just like every other tech conference in the eighteen months since ChatGPT stormed into the world — Atlassian chose to open the event with a discussion featuring AI luminary Dr Fei-Fei Li. She arguably is the person who triggered the outburst of AI innovation that ultimately led to the large language models that power generative AI today, when she created Imagenet, an Internet-scale large dataset that in 2012 was used to demonstrate the power of deep learning models trained on big data.

Her take on how we adapt to the AI age resonated strongly with this tech-savvy audience, drawing several bouts of energetic applause. Here are some of the key points from Li, who for a while was Chief Scientist at Google and is Sequoia Capital Professor of Computer Science at Stanford University and co-director of the Stanford Institute for Human-Centered AI. She was joined in the discussion by Zanny Minton Beddoes, Editor-in-chief of the Economist, along with Mike Cannon-Brookes, Atlassian's co-CEO. It wasn't long before the discussion turned to the issue of the impact of AI on employment and the fear people have that the technology will take their jobs. While acknowledging this fear, Li said that humans are still capable of expression and creativity that technology can't emulate. She went on:

We should have confidence about our human agency, our human creativity and our humanity, with ourselves and with each other. With that confidence, we can use tools — we can use graphics tools, we can use intelligent tools, we can use productivity tools — but fundamentally, they're here to augment and enhance our humanity. We will be creative enough to either create new jobs, or update existing jobs, to use these tools better. But I fundamentally believe our humanity is not going anywhere, and our creativity is more profound than we give ourselves credit for.

The skills therefore that are going to be important in the AI age are these innately human qualities. Beddoes suggested that perhaps the education system should focus more on "the skill set of 'how to be human'." Li expanded on this theme:

There's things that are timeless and fundamental, and they don't change whether you're in the electricity age, or PC age, or AI age. There are other things that will change and we have to adapt to it.

The things that don't change are our core humanity — our ethics, our integrity, our compassion, our creativity, our bonds to each other — these don't change, and we must have them and we must help our children to have them.

The things that change are the skill sets, are the knowledge, are the change in business, technology landscape, the change in society. There, we need people to continue to be open-minded, to listen to each other, to adapt and to learn, and to have growth biceps.

These are the things that, these two buckets ... we have to have them — the core value bucket, and the ability to be adaptive and flexible bucket.

Therefore it's imperative for the education system to adapt rapidly to a new reality — that simply absorbing knowledge isn't enough to set young people up for the rest of their life. She went on:

I imagine that a hundred years later, when the future humans write the history of this third decade of the 21st century, my bet will be the most profound implication of this wave of generative technology is the impact on education. We spend 12 to 16 years of the most precious years of our little humans' lives to achieve at a point that they're evaluated by taking standardized tests that ChatGPT can pass, pretty much today, instantaneously.

Think about that human capital waste, taking that space in all of our brain in the early formative years. I think, given today, ChatGPT or whatever, GPT-4, GPT-5, can pass pretty much every single graduate entrance test, or even the bar tests and all that, we really need to profoundly change our education system.

Where AI will have impact

One area that Li believes will be dramatically improved by the latest AI innovations is science and research, where AI can remove much of the hard graft that's been necessary to make progress in the past. She said:

One area of productivity we're going see — I hope and I'm confident it will explode — is scientific discovery. Scientific discovery actually has been a very meticulous, in general labor-intensive, slow process, because to discover that gene, to create that new material, to find out how to track or whatever sperm whales, all of these are actually really hard mechanics. But I think AI will increase productivity and accelerate these things.

AI can also contribute to creative endeavors. Beddoes revealed that the Economist's designers already use generative AI to help come up with ideas for its cover designs, because they have found it makes them more productive. Echoing a sentiment that's also been expressed by Sam Altman, CEO of Open AI, Li commented that the hallucinations often produced by generative AI aren't necessarily a bug:

Statistical models have uncertain behaviors... With these LLM models, we get into the land of hallucinations. That's one important artefact of this piece of technology today that business leaders and product thinkers have to struggle with. How do we deal with this? Is this a feature or a bug? Sometimes it probably is a feature — that pleasant, magical surprise that machines can give you in, you know, 'Design me a magical animal.' But sometimes it is a bug, if you ask for medical advice. So you have to deal with that.

When Cannon-Brookes suggested that there's a difference between creative imagination and hard facts, she made an intriguing counter-point:

Sometimes imagination turns into fact. Think E=mc2. That kind of imagination gets you closer to truth. So imagination is an incredible thing. It's very profound.

As to the future of AI, she said that she hopes that the focus will move beyond language to more direct representations of the world that humans experience. She said:

I'm very, very excited to go beyond language, because language is the most lossy representation of our world. We open our eyes, we hug the people we love, we take care of the patch of garden we love. Our interaction, our relationship with this world is so much more profound than just the syllabus of words. There's so much more AI technology can do to help us to interact with the world, interact with each other, beyond the form of language, snd that's what I'm excited about.

And I hope that application will really transform especially healthcare and education. I think these are the two areas of our world, of our society, that is so profoundly important to human dignity and human agency. I think we need all of our technology to help lift each other through these industries and applications.

My take

Uplifting, human-centric sentiments from one of the people most associated with the current explosion in AI technology, reminding us of what keeps us distinct from the machines we create.

Loading
A grey colored placeholder image