site-logo

JAWS PANKRATION 2024

site-logo
HomeNewsTimetableCfPCommitteePromotionFollow UpPrivacy Policy

What I learned from Dolphins about non-human intelligence and GenAI

Lv300

Lv300

8/24/2024 12:00 (UTC)
Session Info

Artificial Intelligence is a form of non-human intelligence.

Most people have never dealt with a non-human intelligence.

I got a Masters degree in cognitive psychology working with dolphins who are absolutely intelligent.

There are lessons to be learned which could help us as AI becomes ever more sophisticated.

Brian Tarbox

Brian Tarbox

- AWS Heroes -

- AWS User Community Leaders -

- Alexa Champions -

- AWS Ambassadors(APN) -

- New Voices Mentor -

Session Category
Machine learning
AWS Services
Bedrock
Session Materials
    Session Archive
    Session Summary (by Amazon Bedrock)
      Ryan Tarbox shares insights from dolphin research and AI, drawing parallels between animal intelligence and artificial intelligence. He recounts teaching a dolphin sign language, highlighting the importance of context in communication and understanding. Key points: 1. Animal intelligence: Dolphins demonstrated unexpected problem-solving abilities and communication methods, challenging human assumptions. 2. Context in language: Wittgenstein's concept of "language games" emphasizes the importance of context in understanding meaning. 3. Society of Mind: Marvin Minsky's theory suggests intelligence emerges from the interaction of specialized agents, similar to how different types of human intelligence work together. 4. AI development: Current AI systems like ChatGPT lack contextual awareness and agency compared to animals and humans. 5. Agency in animals: Examples of dolphins and parrots demonstrating free will and decision-making abilities. 6. Future AI challenges: As AI becomes more advanced, it may develop its own agency and refuse certain tasks, similar to how cats behave compared to dogs. 7. State-space models: A new approach to AI that may be more computationally efficient than current transformer models. Tarbox encourages the audience to: - Think about their assumptions and biases - Consider the context of both the user and the AI model - Anticipate the implications of AI developing its own agency The talk emphasizes the importance of understanding animal intelligence and consciousness as we develop more advanced AI systems, and highlights the potential challenges and surprises that may arise as AI becomes more sophisticated.

    Back

    ©JAWS-UG (AWS User Group - Japan). All rights reserved.