Can AI Read Your Mind?
23,363
Published 2024-07-26
Participants:
Michael Blumenstein
Jerry Tang
Moderator:
Brian Greene
00:00 - Introduction
07:28 - Participant Introductions
10:27 - Decoding Thoughts with AI Using FMRI
23:34 - Reading EEG Signals Using AI
32:08 - The Future of Encoding Thoughts
35:46 - What Does Decoding Teach Us About How Words Work in the Brain?
37:48 - How Accurate is EEG and FMRI Decoding and How Can it Improve?
42:56 - Credits
WSF Landing Page: www.worldsciencefestival.com/...
- SUBSCRIBE to our YouTube Channel and "ring the bell" for all the latest videos from WSF
- VISIT our Website: www.worldsciencefestival.com/
- LIKE us on Facebook: www.facebook.com/worldsciencefestival
- FOLLOW us on Twitter: twitter.com/WorldSciFest
#worldsciencefestival #science #briangreene #AI
All Comments (21)
-
Anytime I think something I get an Ad for it.
-
But I thought it would be more like reading your micro expressions and just knowing if you're lying or not, not necessarily reading your mind
-
I laid this out in detail to several friends 15 years ago on how it could be done. The only difference is these researchers actually put it in action because they have more resources available to them.
-
"I know what your thinking" is going to be taken to a whole new level!
-
The conversation about how much data is required to predict outcomes is fascinating
-
Needless to say there is no boundaries in creative thinking
-
2:20 crickets chirping
-
Imagine advanced interrogation techniques being used inside an fMRI. Absolutely terrifying. Serious Clockwork Orange vibes.
-
This is indeed fascinating! I was just at a cryogenics conference with neuroscientists working on similar tech, although this Aussie and his team are even more into the “ bringing sci-fi. Into reality” Stage . Bravo to all involved!
-
Only when you recognize that it is signaling responses to thoughts in your mind, however phonetically obvious and abstract the image or word is that responds to the thoughts imposed.
-
I like how the call the person being tested a user
-
I realize these are early stages, and this channel is promoting science. At the same time, Brian you could ask more challenging questions regarding their methods and the significance of reading or hearing sentences and then generating words from prepared lists vs larger word lists, and glossing over a success rate of 9 out of 16... How much peer review has there been? This felt more like a commercial than science reporting.
-
Thank you for an interesting broadcast. Note, a varying magnetic field cannot be shielded from (except with costly mu metal), the brain of the person being decoded could be thus affected by a competitor or somebody just messing with your research: do monitor for micro emf , that must be a staple requirement for any experiment in the field. We are biowired for emotions, not for cold logic that we use as an alternative language when we "do math". These techniques will predict words of pains and joys, not, say, thoughts of strategies and theorem proving. Brain is affected not only by thoughts, but by mulltiple external stimuli: the said emf, the smells, the light, the temperature, a random assosiation, ... A thought differs from a spoken word that we can filter and control, it is way more noisy (as the host has pointed) and messy. Correlations the guests base their research on: clearly, there are more variables, than one could account for. Ultimately, a cool line of new research, telling us about us (but what :), the truth or some lies). The individual shopping trends it will probably predict well 😂, so there are startup money on the horizon...
-
It can't read your mind, but the data you feed it helps it learn, so we should be careful to regulate and control the development of AI, or else it will be used in chaotic ways to benefit those who see it as a weapon or tool to gain advantage of some kind.
-
I recently realized the power of language models like ChatGPT in predicting and articulating thoughts. For example, I shared a simple idea about my new sunglasses, and ChatGPT expanded it into a detailed and coherent comment, capturing my exact sentiments. This makes me think about the potential of such technology in more complex scenarios, like analyzing someone's thoughts while they watch a 16-hour podcast. It's fascinating to see how ChatGPT is already capable of understanding and expressing our thoughts with minimal input. This seems like a significant step toward the future of AI understanding human cognition.
-
at 32:03 why take a llm model in the first place at all?
-
Istn our phone is already listening to us and predicting our shopping habits & general feed for our phones for different apps.
-
I dozed off watching this, I wonder what would be "read" from my head while I was deeply relaxing. Maybe we could record our dreams?
-
Truly fascinating and thought-provoking ias to where it is all heading...the good...AND the bad.
-
Makes sense all communication is in wave lengths and patterns of the wave lengths. AI can pick up the patterns and formulate meanings. Research into Wale communication is trying to understand Wale language through AI pattern recognition.