Lifeboat Foundation

Continued thoughts 💭 about humanity and our beloved AI

Maybe interesting.

173K likes, — Philosophy & Literature (@philosophyofexistence) on Instagram: ‘Aristotle’s Paradox of Time’

Researchers at Washington State University have been monitoring challenges honeybees face for nearly 20 years, and they said this year could be one of the worst ones for the important pollinators in decades.

However, they have also been working to create robot bees to help with pollination. KCBS Radio’s Holly Quan spoke with Ryan Bena, a PhD student at the University of Southern California and co-author of the study about the project.

“Essentially we built this this robot – it’s about 95 milligrams,” he explained. “So it’s roughly the size of… an actual insect bee. And we use flapping wings. So for flapping wings to fly and control the bee, you know, fly through the air… what’s unique and sort of interesting about our particular robot is that we finally developed a way to coordinate the flapping of these four wings so that we can control the bee in every direction.”

Paragraphica may resemble a camera, but it does not capture light to generate images. Instead, the device dispenses with a lens and sensor altogether, and relies on AI to produce results.

In recent years, electronics engineers have been trying to develop new brain-inspired hardware that can run artificial intelligence (AI) models more efficiently. While most existing hardware is specialized in either sensing, processing or storing data, some teams have been exploring the possibility of combining these three functionalities in a single device.

Researchers at Xi’an Jiaotong University, the University of Hong Kong and Xi’an University of Science and Technology introduced a new organic transistor that can act as a sensor and processor. This transistor, introduced in a paper published in Nature Electronics, is based on a vertical traverse architecture and a crystalline-amorphous channel that can be selectively doped by ions, allowing it to switch between two reconfigurable modes.

“Conventional artificial intelligence (AI) hardware uses separate systems for data sensing, processing, and memory storage,” Prof. Wei Ma and Prof. Zhongrui Wang, two of the researchers who carried out the study, told Tech Xplore.

The mammalian retina is a complex system consisting out of cones (for color) and rods (for peripheral monochrome) that provide the raw image data which is then processed into successive layers of neurons before this preprocessed data is sent via the optical nerve to the brain’s visual cortex. In order to emulate this system as closely as possible, researchers at Penn State University have created a system that uses perovskite (methylammonium lead bromide, MAPbX₃) RGB photodetectors and a neuromorphic processing algorithm that performs similar processing as the biological retina.

Panchromatic imaging is defined as being ‘sensitive to light of all colors in the visible spectrum’, which in imaging means enhancing the monochromatic (e.g. RGB) channels using panchromatic (intensity, not frequency) data. For the retina this means that the incoming light is not merely used to determine the separate colors, but also the intensity, which is what underlies the wide dynamic range of the Mark I eyeball. In this experiment, layers of these MAPbX₃ (X being Cl, Br, I or combination thereof) perovskites formed stacked RGB sensors.

The output of these sensor layers was then processed in a pretrained convolutional neural network, to generate the final, panchromatic image which could then be used for a wide range of purposes. Some applications noted by the researchers include new types of digital cameras, as well as artificial retinas, limited mostly by how well the perovskite layers scale in resolution, and their longevity, which is a long-standing issue with perovskites. Another possibility raised is that of powering at least part of the system using the energy collected by the perovskite layers, akin to proposed perovskite-based solar panels.

AI is everywhere. Its use is being debated in headlines, on social media and around dinner tables. To some, the rate of AI acceleration is concerning, with many technology leaders calling for a six-month pause in the training of new systems to better understand the impact such tools are having. To others, AI is seen as the cornerstone of the fourth industrial revolution, the latest disruptive technology opening up possibilities for new ways of learning, working and living that we have never experienced before.

Yet, disruptive technologies are nothing new. They have been changing the way we live and work for decades. And these changes have not been without consequences, particularly in the form of economic dislocation and social upheaval. Automation in manufacturing has streamlined mass production and driven down costs; Ecommerce platforms have reshaped the way we shop and do business; even online education has found new ways to provide flexible and affordable ways of learning, delivering opportunities to millions across the globe that simply were not available before.

Presently, much of the discussion around the impact of AI is based on conjecture. However, it is widely agreed that it will have a major impact on jobs and even has the potential to call into question the very fundamentals of what work is. What is not understood is how AI will play out across society in the longer term. Will it, like previous technological revolutions, deliver short-term disruptions followed by long-term benefits, or will it be the catalyst for new ways of learning and upskilling and help reduce the widening digital divide?

AI researchers have built a Minecraft bot that can explore and expand its capabilities in the game’s open world — but unlike other bots, this one basically wrote its own code through trial and error and lots of GPT-4 queries.

Called Voyager, this experimental system is an example of an “embodied agent,” an AI that can move and act freely and purposefully in a simulated or real environment. Personal assistant type AIs and chatbots don’t have to actually do stuff, let alone navigate a complex world to get that stuff done. But that’s exactly what a household robot might be expected to do in the future, so there’s lots of research into how they might do that.

Minecraft is a good place to test such things because it’s a very (very) approximate representation of the real world, with simple and straightforward rules and physics, but it’s also complex and open enough that there’s lots to accomplish or try. Purpose-built simulators are great, too, but they have their own limitations.

Stepping closer to AI regulation.

As AI continues to develop at a rapid rate, concerns have been raised by experts and the so-called AI ‘godfathers’ about the imminent risks it could pose to people’s privacy, human rights, and safety.

Amid steps taken by the United Kingdom government and the European Union, in partnership with the U.S, to regulate the technology, a member of the non-statutory AI Council of the U.K. government has said that the very powerful artificial general intelligence (AGI) systems may eventually have to be banned.

Midcap information technology (IT) services firm, Mphasis, on Thursday, announced the setting up of a business unit dedicated to generative artificial intelligence (AI). The unit will offer advisory on adoption of generative AI solutions to clients, develop the company’s own generative AI properties, offer licenses to over 250 AI models through the company’s ‘Hyperscaler’ solutions marketplace for clients, partnerships with 50 startups for helping clients build solutions, and offer conversational AI tools such as chatbots for clients to deploy in their business.

Anup Nair, who has so far served as senior vice-president and chief technology officer (CTO) of Mphasis Digital, will helm the unit, called Mphasis.ai, as its chief architect and CTO.

The launch of the dedicated business unit comes after a flurry of similar launches by pretty much every large-cap IT services firm in the country. On 6 April, Tech Mahindra became the first of the large domestic IT firms to launch a generative AI solution, called ‘Generative AI Studio’, to help clients deploy the technology for content generation use cases.