“AI is the most important thing humanity is working on today."
-Jeff Hinton
AI is a game changer, full of new opportunities and risks. How can we take advantage of the opportunities, simple and massive? How can we predict and minimize the risks, from the mundane to the mind-numbing?
On November 9, 2023, 1Flourish Capital hosted experts from leading AI startups, Meta, Google, Apple, and venture capital firms to explore the challenges and opportunities of responsible AI.
AI is exploding human productivity
Through AI, we can quickly automate tiresome tasks to be completed with greater speed and higher precision, while at a lower cost. Tamas Cser, co-founder of Functionize, shared how Functionize makes software testers up to ten times more productive. Projects that test tens of thousands of views and functions across global geo locations were not feasible before with manual coding, but now can be achieved with AI in hours. Tamas says his own productivity has grown massively with code writing tools like co-pilot. One weekend, he translated a 4000-line Python program to go with the help of AI.
AI can augment human creativity and democratize creativity
Sherol, an AI Research Engineer from a top tech firm, has been working on ways that AI can provide creative leverage for over 10 years. She works on representational alignment for how generative models can better adapt to user intent, and is excited about the ways we can use technology to teach and cultivate creativity in people across diverse communities.
Andrew Laffoon, CEO of Mixbook, has used machine learning for over a decade to help people turn photo libraries into meaningful memories in photobooks. Andrew said that many users get stuck in the process of grouping and naming their photo collections. After recently launching a feature that uses computer vision and LLMs to suggest photo captions, a customer trying to finish a book late at night for a family event said it helped her complete it in minutes instead of days and get the photobook memories delivered just in time for an important family event.
AI opens new areas of attack
Bob Kruse, a security growth lead at Arctic Wolf, warned that AI creates new spaces for bad actors to attack our personal devices, cloud services, and more. The number of vulnerabilities discovered and used without us knowing, or zero-day attacks, is rapidly increasing. Bob strongly recommends that we accept that some people will use AI for harm. We should prepare and monitor for attacks using leading tools such as Revelstoke, which he founded. We should also ensure that AI systems are transparent and accountable.
Will AI be sentient soon?
Tamas quickly noted that it depends on how we define sentient. Machines now learn logic without human supervision and often from data that is AI generated. The size and scale of inputs data and model parameters lead to new forms of logic that the system creators might not understand. AI already hallucinates frequently, showing that AI doesn't have to be sentient as humans in order to arrive at new ideas and conclusions. If AI is used to automate tasks and improve decision-making, we have to be careful that it can’t hallucinate and automatically implement bad decisions.
What does it mean to be human? What does it mean to be conscious?
Sherol is often asked how we should be using AI. She feels that the question is ultimately about the type of people we want to be as part of human society. That answer should inform how we use the tools we invent, not the other way around.
Let’s work together to drive productivity and creativity that lead to a better future. I hope we are intentional and principled. AI has the potential to solve many of the world's biggest problems and create a better future for everyone. However, we must be on the lookout for bad actors.
Prepare. Not everything you read on the Internet is true.
Comments