Lifeboat Foundation

The bulk of the computing in state-of-the-art neural networks comprises linear operations, e.g., matrix-vector multiplications and convolutions. Linear operations can also play an important role in cryptography. While dedicated processors such as GPUs and TPUs are available for performing highly parallel linear operations, these devices are power-hungry, and the low bandwidth of electronics still limits their operation speed. Optics is better suited for such operations because of its inherent parallelism and large bandwidth and computation speed.

Built from a set of spatially engineered thin surfaces, diffractive deep neural networks (D²NN), also known as diffractive networks, form a recently emerging optical computing architecture capable of performing computational tasks passively at the speed of light propagation through an ultra-thin volume.

These task-specific all-optical computers are designed digitally through learning of the spatial features of their constituent diffractive surfaces. Following this one-time design process, the optimized surfaces are fabricated and assembled to form the physical hardware of the diffractive optical network.

In this clip from our 2017 event titled ‘Evolution of the Mind, Consciousness and AI,’ the esteemed philosopher and cognitive scientist, Daniel Dennett is joined by a group of panellists to explore how much we understand about the human mind, and what the creation of artificial consciousness means for our future. Watch and let us know in the comments if you think Dennett’s theories still hold true in light of the rapid developments in AI since he joined us.

See the full session here: • Daniel Dennett on the Evolution of th…

Continue reading “Consciousness, AI and the Future of Humanity” »

Is an American philosopher, writer, and cognitive scientist whose research centers on the philosophy of mind, philosophy of science, and philosophy of biology, particularly as those fields relate to evolutionary biology and cognitive science.

Recorded, Nov 22, 2017.

The study, authored by five MIT researchers and titled Beyond AI Exposure, delves deep into the practicalities of replacing human labor with AI in the US, focusing on tasks that lend themselves to computer vision, such as those performed by teachers, property appraisers, and bakers.

Like many of us, you might find yourself nodding to a familiar digital doomsday chorus that vibrates through offices and coffee shops alike: AI will take my job!

Continue reading “Rethinking AI’s impact: Study reveals economic limits to job automation” »

MachineLearning clinical prediction models fail to generalize across trial data, a new Science study finds. The results “require reexamination of the practical challenges that precision medicine is facing.” Learn more in a new Science Perspective:

The prediction of individual treatment responses with machine learning faces hurdles.

Frederike H. Petzschner [email protected] Authors Info & Affiliations

Continue reading “Practical challenges for precision medicine” »

A team of MIT researchers has found that in many instances, replacing human workers with AI is still more expensive than sticking with the people, a conclusion that flies in the face of current fears over the technology taking our jobs.

As detailed in a new paper, the team examined the cost-effectiveness of 1,000 “visual inspection” tasks across 800 occupations, such as inspecting food to see whether it’s gone bad. They discovered that just 23 percent of workers’ total wages “would be attractive to automate,” mainly because of the “large upfront costs of AI systems” — and that’s if the automatable tasks could even “be separated from other parts” of the jobs.

That said, they admit, those economics may well change over time.

Google Deepmind says that a new artificial intelligence system has made a major breakthrough in one of the most difficult tests for AI.

The company says that it has created a new AI system that can solve geometry problems at the level of the very top high-school students.

Geometry is one of the oldest branches of mathematics, but has proven particularly difficult for AI systems to work with. It has been difficult to train them because of a lack of data, and succeeding requires building a system that can take on difficult logical challenges.

The article repeats itself a bit but there’s some good parts about an exoskeleton, advanced algorithm and bipedal robots and prosthetics. It’ll basically apply to those future industries.

We typically don’t think about it whilst doing it, but walking is a complicated task. Controlled by our nervous system, our bones, joints, muscles, tendons, ligaments and other connective tissues (i.e., the musculoskeletal system) must move in coordination and respond to unexpected changes or disturbances at varying speeds in a highly efficient manner. Replicating this in robotic technologies is no small feat.

Now, a research group from Tohoku University Graduate School of Engineering has replicated human-like variable speed walking using a musculoskeletal model – one steered by a reflex control method reflective of the human nervous system. This breakthrough in biomechanics and robotics sets a new benchmark in understanding human movement and paves the way for innovative robotic technologies.

Last week, Facebook founder Mark Zuckerberg announced that he’s going to purchase hundreds of thousands of expensive AI processing chips — and experts are mighty worried about what he plans to use them for.

In the same Instagram post announcing his planned purchase of 350,000 Nvidia’s H100 graphics chips, which average about $30,000 apiece and are considered the gold standard for powering AI models, Zuckerberg said that he wants to build an open-source artificial AGI, the industry term for the point at which AI reaches or even surpasses human-level intelligence.

While there’s still an open debate about whether AGI is even possible, the prospect itself is enough to give some researchers pause.