Think and you’ll miss it: researchers at MIT claim to have successfully created analog synapses that are one million times faster than those in our human brains.

Just as digital processors need transistors, analog ones need programmable resistors. Once put into the right configuration, these resistors can be used to create a network of analog synapses and neurons, according to a press release.

These analog synapses aren’t just ultra-fast, they’re remarkably efficient, too. And that’s pretty important, because as digital neural networks grow more advanced and powerful, they require more and more energy, increasing their carbon footprint considerably.

A research team has demonstrated that analog hardware using ECRAM devices can maximize the computational performance of artificial intelligence, showcasing its potential for commercialization. Their research has been published in Science Advances.

The rapid advancement of AI technology, including applications like generative AI, has pushed the scalability of existing digital hardware (CPUs, GPUs, ASICs, etc.) to its limits. Consequently, there is active research into analog hardware specialized for AI computation.

Analog hardware adjusts the resistance of semiconductors based on external voltage or current and utilizes a cross-point array structure with vertically crossed memory devices to process AI computation in parallel. Although it offers advantages over digital hardware for specific computational tasks and continuous data processing, meeting the diverse requirements for computational learning and inference remains challenging.

Google’s Gemini 1.5 Pro revolutionizes AI with top LMSYS scores, outperforming GPT-4 and Claude-3.5, promising transformative impact across industries while raising ethical questions.

This is a temporary transcription of Josha Bach’s talk on artificial intelligence philosophy at CCC 2023.

More informations here: