Neural Notes 2: Our Updates + Geoffrey Hinton's Take on Digital Intelligence 🤖

Hello Everyone!

Your favourite co-presidents are back with some new Neural Notes 📝

What have we been up to?

Our conference planning is officially in full swing, and we’ve got some exciting updates to share. Over the past couple weeks, we’ve been scouting out the coolest conference venues in town, on the hunt for the perfect space where ideas are shared, connections are made, and the future of AI takes center stage. We’ve been chatting with event managers and brainstorming ideas that fit our ambitious plans to expand our AI presence in Canada.

So get ready to mark your calendars in 2024 for an unforgettable event.

ProjectX is Underway:

ProjectX is a four-month long machine learning research competition hosted annually by UofT AI. The competition has already begun with top schools in North America competing to propose the best research solution in the area of computational efficiency. We’re also looking for more mentors to guide our ProjectX teams in heavy research areas. If you’re an AI expert with a passion for shaping the future generation of innovators, we’d love to hear from you.

More information about ProjectX 2023: here

We’re partnering with Vector Institute for yet another event!

We’re excited to announce our “Transforming Healthcare Research Through AI: Empowering the Future Generation” seminar in collaboration with the Undergraduate Research Students’ Association (URSA), happening November 15th. For this event, we’ve partnered with the Vector Institute to explore the dynamic relationship between healthcare and AI through talks from top-notch speakers. This seminar is your chance to connect with professionals, researchers, and experts who are shaping the future of AI and healthcare. Keep in touch with us through our Instagram (@uoftaigroup) and stay tuned for registration forms that will be posted on there so you can attend the event.

Your Burst of AI News: Geoffrey Hinton’s take on Digital Intelligence 🧠

This past Friday, the UofT AI team attended an enlightening conversation by the renowned Geoffrey Hinton and Vector Institute. Hinton delved deep into the foundational differences between digital computers and their biological counterparts.

Here are some key takeaways:

• Digital computers were initially conceived to allow humans to dictate their exact functions. Their design demands meticulous fabrication and substantial energy consumption. However, they guarantee that the same model can function identically across diverse hardware, ensuring the model's immortality. As we transition to computers capable of learning autonomously, Hinton proposed a radical idea: forgoing the traditional separation of software from hardware.

• Yet, this approach poses challenges. Leveraging the distinct analog properties of hardware implies a computational mortality. If the hardware succumbs, so does its learned knowledge. While it's feasible to transfer this knowledge to a younger analog device by having it imitate the older device's outputs, the "education" is laborious and time-consuming.

• In contrast, digital computation presents unparalleled advantages. It permits the seamless operation of multiple identical models across varied hardware. Such capability enables thousands of identical digital entities to scrutinize diverse datasets, subsequently sharing their learnings through efficient weight change averaging. It's this remarkable capability that empowers chatbots like GPT-4 and Gemini to amass knowledge surpassing any human. Digital computation also supports the back propagation learning procedure, which scales significantly better than methods tailored for analog devices.

• Hinton concluded with a profound foresight: digital computation, with its expansive knowledge acquisition capabilities, might soon overshadow biological computation, potentially becoming more intelligent than humans. The eternal nature of digital entities and their evolutionary detachment could render them less prone to the pitfalls of religion and conflicts.

Get Involved with Uoft AI

As always we’re on the lookout for partners, speakers, and sponsors who want to take part in our annual conference or partner with us for events throughout the year. If you or someone you know is as passionate about the future of AI as we are, shoot us an email at [email protected] or just reply to this thread!

Let’s make this event bigger and better together!

Cheers,

Your Presidents,

Jai and Alexis