IBM

Will neuromorphic computers accelerate AGI development?

Will neuromorphic computers accelerate AGI development?

#neuromorphic #computers #accelerate #AGI #development

“Dr Waku”

Neuromorphic computers are devices that try to achieve reasoning capability by emulating a human brain. They are a different type of computer architecture that copies the physical characteristics and design principles of biological nervous systems. Although neuromorphic computations can be…

source

 

To see the full content, share this page by clicking one of the buttons below

Related Articles

35 Comments

  1. Thank you to the viewer who asked for a neuromorphic computer video! I think it it ended up being an interesting topic.

    Don't forget to join my discord and support me on patreon if you like. Also, sorry this video is two days late from my normal Sunday publishing time.

  2. 📝 Summary of Key Points:

    📌 Neuromorphic computing involves creating computers that replicate the physical characteristics and design principles of the human brain. It aims to closely mimic the brain's biology, unlike current AI algorithms like deep learning.

    🧐 The human brain is incredibly complex, with billions of neurons and trillions of synapses. Current supercomputers have significantly lower processing capabilities compared to the brain. Neuromorphic supercomputers, like Deep South, aim to emulate the brain's synaptic operations and are built on flexible and customizable platforms like FPGAs.

    🚀 Neuromorphic computing could have a significant impact on the development of artificial general intelligence (AGI). If large language models face limitations, neuromorphic research could identify breakthroughs. Additionally, neuromorphic computers could offer more efficient and specialized architectures, reducing energy consumption compared to GPU implementations.

    💡 Additional Insights and Observations:

    💬 "Neuromorphic systems aim to replicate the brain's functioning more closely than traditional neural networks, which are an abstraction of the real thing."

    📊 The human brain consumes about 20 watts of power, while supercomputers require much more energy.

    🌐 Deep South, a neuromorphic supercomputer, aims to emulate 228 trillion synaptic operations per second, close to the number of synaptic connections in a human brain.

    💬 "If large language models are sufficient to achieve AGI, there may be no need for new architectures."

    📊 By 2027, AI servers using GPUs could consume about 0.5% of the world's total energy consumption.

    📣 Concluding Remarks:

    Neuromorphic computing holds promise for creating computers that closely resemble the human brain. The development of neuromorphic supercomputers, like Deep South, could advance AGI research. These computers, built on flexible platforms like FPGAs, offer the potential for more efficient and specialized architectures, reducing energy consumption compared to GPU implementations.
    Generated using TalkBud

  3. Deepsouth's 228 trillion synaptic operations per second is 1000 times slower than a human brain 🙁 The human brain has several hundreds of trillions of synapses and every synapse can conduct spikes 1000 times per second

  4. Somehow mimicking human brains just doesn't sound like the best road map to follow on this. Arent we after something better? Why start with a duplicate of us? Record isn't so great.

  5. You didn't excellent job on this video. There is some missing information though. Deep South is reportedly going to run off of only 30 watts which is remarkably close to the human brain.

    RainAI is the company making the chips for deep south. This is the company that open AI and Sam Altman himself are highly invested in. Open AI is scheduled to receive these same chips and a computer of the same size in October.

    This investment is what triggered the firing of Sam Altman last year. The stated reason was that it posed an existential risk to humanity. The program that is to run on this computer called Quallia or Q* has shown the ability to reprogram itself and is based off of two forms of self-awareness programming.

    Sam Altman himself called all of this information, unfortunately leaks and not rumors or lies.

    It is also very interesting that 2 days after this AI reprogrammed itself 60,000 times and and broken cryption with mathematic proofs that it came up with itself The department of Justice showed up. Now we are striking terrorist cells that are hidden all over the world as if we just discovered their locations.

    There are multiple other companies working on NPUs. Last month a very interesting paper was released on how actual brain cells grown on a computer chip. We're able to decipher human language and work as a voice recognition program. It was able to distinguish between two different human voices.

    I love what you're doing with the channel.

  6. We really are at the tipping point of human technology. Once our AIs become intelligent enough to make discoveries of their own, our technology will far surpass what we have now in ways we can't imagine.

  7. I’ve been following your videos since the “Why no one saw ChatGPT coming” video. I absolutely love how every video you make is packed with information as well as how you announce the organization of topics at the beginning of each video. It helps me organize my thoughts as I hear you talk. Keep up the excellent work!

  8. I think not many years from now AI will probably consume more than 50% of energy of Earth and beyond. It shouldn't take too long to expand into space for energy needs. While we actively start to use fusion energy.

  9. This is fascinating. I'm intrigued by the first BCI implant! How fast will the field explode? Could brain-cloud links accelerate AGI? Seems less scary than lab-grown brains (eek, ethics!). ☁

  10. Lots of people predicted that one superhuman artificial general intelligence would take over the world, maybe Google, maybe Openai, maybe the NSA, maybe the Chinese or Japanese, but no one predicted it would be Australia.

Leave a Reply