I got back from the picturesque city of Vancouver a couple of weeks ago, after attending TED 2023 (the main conference) for the second time. Apart from being one of the best events from a production point of view (lots of attention to detail from communication to infrastructure and design) it was an opportunity to learn, network and get inspired. The theme of this year’s conference was ‘Possibility’ and there was a significant focus on AI and generative AI, and the possibilities they unlock.
It’s hard to summarise a whole week in a short blog post, but I wanted to share a few thoughts on key themes from the conference and what I learned.
Theme 1: The Power of Generative AI
The first AI talk was by Greg Brockman, co-founder and CEO of OpenAI. It was a glimpse and live demo of some of the features coming to ChatGPT in the next few months including the plugins integration, that connects ChatGPT to the Internet, data visualisation and graph creation with code interpreter and more. I recommend you watch it (video of the talk embedded below).
Perhaps more interesting was the Q&A that followed with Chris Anderson, where Greg was asked why did they unleash such a powerful tool on the world without taking into account the many implications it will have on society and also should AI development be slowed down?
Greg’s response (and Sam Altman’s as well in the video below) is a window to OpenAI’s philosophy on this. In a nutshell, they believe that AI development is inevitable, and that it cannot be locked in a lab to be developed in secret. They believe that having the product in the hands of users where the ‘rubber hits the road’ provides them with valuable feedback on how the products should be improved.
Theme 2: AI in Science
Karen Bakker, a professor at the University of British Columbia and conservation technology researcher, gave a talk on how AI is enabling us to capture the world of ultra or infra sonic sounds, which animals use to communicate with one another, and start to decipher what some of those sounds might mean. It was a thought provoking session as it might be possible to achieve interspecies communication as we continue to decode how Orcas, dolphins, bats, etc talk to one another. Using generative AI, scientists were able to reproduce these sounds to try to communicate with bees, for example, but so far with little success.
Scientists are even able to translate some variations of animal speech, while generative AI is able to imitate some of these sounds, allowing us to communicate with nature like never before – and bringing along some difficult challenges, too.
Can you imagine a Dolphin giving a TED talk about the impact of warming oceans in a few years? Karen goes deeper on this topic in her book, The Sounds of Life.
Theme 3: AI in Education
When it comes to ChatGPT and generative AI, you can be certain that students of all ages are early adopters, as it makes it much easier for them to turn in their homework 😉
TED featured two interesting examples of AI in education that showed the potential of generative AI to give every teacher an AI teaching assistant and a personal tutor for every student. The first one was Sal Khan, founder of Khan Academy. Sal demonstrated Khanmigo, a GPT-4 powered teaching assistant.
The second was a talk by the Luis von Ahn, CEO of DuoLingo, the most popular language learning school, which started offering a new AI-powered language tutor to consumers, powered by GPT-4. The interesting part was their AI-powered gamification to motivate users to learn. For example, DuoLingo figured out that the best time to send users notifications needs to be personalised. They are most likely to be available for a session at the same time they opened the app in their previous session. In addition, they communicate with users in a very natural language and increase engagement as a result. Here’s an interesting article on what DuoLingo does differently to increase retention.
Theme 4: the risks and limitations of AI
Overall, TED did a good job balancing the promise of AI with the dangers it poses and its limitations. Computer scientist Yejin Choi spoke about the importance of giving AI human values. In her view, many AI systems brings three big issues with it:
- AI models are expensive to train,
- Their power is concentrated to only a few tech companies and
- The environmental impact is massive
Gary Marcus spoke about the risk AI poses on spreading disinformation and called for governments to regulate this technology before it’s too late. Perhaps the most controversial talk was by Eliezer Yudkowsky, who was invited to speak only a week before the conference and gave a 6 minutes talk that he read from his phone. His message was a scary one: superintelligent AI could probably kill us all and by the time we realise it, it will be too late. Eliezer recently recorded a long podcast episode with Lex Fridman if you want to learn more about his views. Also, Tom Graham, the founder and CEO of Metaphysic, the company behind deep fake Tom Cruise, talked about new opportunities (and risks) for this technology in media and entertainment. Other risks that were discussed: AI weapons, fake news, copyright infringement, bad actors leveraging AI agents… and so on. No real answers to these risks, apart from a sense that the industry should agree on an ethical code for AI and help regulators set some boundaries.
Final thoughts
There was much more than AI discussed, and I would be remiss not to mention the work of some inspiring non-profits that presented, talking about issues of social justice, racial inequality, climate or the member of Pussy Riot that served two years in prison and is wanted by Russia. There were also the classic motivational speakers that sometimes made me move uncomfortably in the chair, like I’m in the SNL version of TED, but generally speaking I was more attracted to the talks that had a real message and story to tell, vs. great delivery/theatrics. Some of the most impactful talks had nothing to do with AI – a death dula that accompanied hundreds of people and their families to help them prepare for their death, is possibly one of my favourites and made me think about my own mortality. Another talk that landed was about the importance of supporting boys (not at the expense of girls, but the balance swung too far the other way). I encourage you to watch them all on Youtube when they get published.
Overall, the talks were only part of the experience. Much of my learning came from conversations with interesting people, including Reed Hastings, founder of Netflix, Yat Siu, co-founder and chairman of Animoca, and many other professors, entrepreneurs, VCs and philanthropists. I ran into friends, ex colleagues from Google and made new friends.The TED community is what genuinely makes TED special.
TED made me put our work at Remagine Ventures in context, and realise that despite all the shit we hear in the news, there’s a lot of progress being made that doesn’t normally get the spotlight. It was also a good reminder that the information we get used to consume, and what you will read in the news is not necessarily what’s most important, so stay curious and keep on learning. We are living through a time of unprecedented progress in science and technology, but despite all that tech (or maybe because of it) there’s no replacement to human connection and interaction. Thank you for reading.
I’m a former general partner at google ventures, head of Google for Entrepreneurs in Europe and founding head of Campus London, Google’s first physical hub for startups.
I’m also the founder of Techbikers, a non-profit bringing together the startup ecosystem on cycling challenges in support of Room to Read. Since inception in 2012 we’ve built 11 schools and 50 libraries in the developing world.
Related
