News Archive - Page 45 of 67

Making AI smarter: a multisensory, integrated artificial neuron

Sensors in robots and other devices typically run separately. However, “allowing sensors to communicate directly with each other can be more efficient in terms of energy and speed,” emulating the human brain, says Saptarshi Das, associate professor of engineering science and mechanics at Penn State.

Penn State researchers applied this synergistic approach in developing the first multisensory integrated artificial neuron, which they announced today (Sept. 15) in the open-access journal Nature Communications.

Creating a tactile-light neuron

The team focused on integrating a tactile sensor and a visual sensor. This combination allows the output of one sensor to modify the other, with the help of visual memory. (For example, a flash of light could enhance the chance of successful movement through a dark room.)

The researchers fabricated the multisensory neuron by connecting a tactile sensor to a phototransistor, based on a monolayer of molybdenum disulfide, a compound that exhibits unique electrical and optical characteristics, useful for detecting light and supporting transistors.

The sensor generates electrical spikes in a manner reminiscent of neurons processing information, allowing it to integrate both visual and tactile cues.

*Visual and tactile stimuli are integrated in this bio-inspired artificial neuron (credit: Tyler Henderson/Penn State)*

Simulations

To simulate touch, the tactile sensor used the triboelectric effect, in which two layers slide against one another to produce electricity (meaning the touch stimuli were encoded into electrical impulses). To simulate visual input, the researchers shined a light into a monolayer molybdenum disulfide photo memtransistor (which can remember visual input), similar to how a person can hold onto the general layout of a room after a quick flash illuminates it.

They found that the sensory response of the neuron—simulated as electrical output—increased when both visual and tactile signals were weak.

Das explained that an artificial multisensory neuron system could enhance sensor technology’s efficiency, paving the way for more eco-friendly AI uses. As a result, robots, drones and self-driving vehicles could navigate their environment more effectively while using less energy.

Combining sensors to mimic how our brains actually work

“Biology enables small organisms to thrive in environments with limited resources, minimizing energy consumption in the process,” said Das, who is also affiliated with the Materials Research Institute.

“The requirements for different sensors are based on the context. In a dark forest, you’d rely more on listening than seeing, but we don’t make decisions based on just one sense,” he noted. “We have a complete sense of our surroundings, and our decision-making is based on the integration of what we’re seeing, hearing, touching, smelling, etc.

“The senses evolved together in biology, but separately in AI. In this work, we’re looking to combine sensors and mimic how our brains actually work.”

The Army Research Office and the National Science Foundation supported this work.

Citation: Sadaf, M. U., Sakib, N. U., Pannone, A., Ravichandran, H., & Das, S. (2023). A bio-inspired visuotactile neuron for multisensory integration. Nature Communications, 14(1), 1-12. https://doi.org/10.1038/s41467-023-40686-z (open-access)

Let us know your thoughts! Sign up for a Mindplex account now, join our Telegram, or follow us on Twitter.

Your brain’s amygdala may release cannabinoid molecules during stress, researchers find

When you’re under stress, your brain’s amygdala (a key emotional brain center) may release cannabinoid molecules*, activating the same brain receptors that THC** does, Northwestern Medicine researchers have discovered.

These endogenous (body’s own) feel-good cannabinoid molecules dampen incoming stress alarms from the hippocampus, a memory and emotion center in the brain, the researchers report in a study with mice, published Sept. 12 in the journal Cell Reports.

Overcoming stress disorders: new drug development

Stress heightens risk for many psychiatric disorders. This new finding opens a new avenue for drug development to treat psychiatric disorders from conditions like generalized anxiety, major depression and post-traumatic stress disorder (PTSD), the researchers suggest.

“Understanding how the brain adapts to stress at the molecular, cellular and circuit level could provide critical insight into how stress is translated into mood disorders and may reveal novel therapeutic targets for the treatment of stress-related disorders,” said study author Dr. Sachi Patel, chair of psychiatry and behavioral sciences at Northwestern University Feinberg School of Medicine and a Northwestern Medicine psychiatrist.

“The endocannabinoid system is one of the leading signaling systems that has been identified as a prominent drug-development candidate for stress-related psychiatric disorders, Patel said.

Removing cannabinoid receptor decreases pleasure and ability to cope

When these scientists removed the target of these cannabinoids, the cannabinoid receptor type 1, it resulted in poorer ability to cope with stress and motivational deficits in the mice. Specifically, when the receptor target of these endogenous cannabinoids was removed at hippocampal-amygdala synapses, mice adopted more passive and immobile responses to stress and had a lower preference to drink a sweetened sucrose water after stress exposure.

This finding may relate to anhedonia (decrease in pleasure), often experienced by patients with stress-related disorders such as depression and PTSD.

“Determining whether increasing the levels of endogenous cannabinoids can be used as potential therapeutics for stress-related disorders is a next logical step from this study and our previous work,” said Patel. “Also, there are ongoing clinical trials in this area that may be able to answer this question in the near future.”

* Cannabinoids are a class of biological compounds that bind to cannabinoid receptors. They are most frequently sourced from and associated with the plants of the Cannabis genus, including Cannabis sativa, Cannabis indica, and Cannabis ruderalis.

** THC (tetrahydrocannabinol) is the substance that’s primarily responsible for the effects of marijuana on a person’s mental state.

Citation: Kondev, V., Najeed, M., Yasmin, F., Morgan, A., Loomba, N., Johnson, K., … & Patel, S. Sept.12, 2023. Endocannabinoid Release at Ventral Hippocampal-Amygdala Synapses Regulates Stress-Induced Behavioral Adaptation. Cell Reports. https://papers.ssrn.com/sol3/papers.cfm?abstract_id=433

Let us know your thoughts! Sign up for a Mindplex account now, join our Telegram, or follow us on Twitter.

‘Lightning’ system combines light and electrons to unlock faster, greener computing for machine-learning models

The growing demand for high-performance computers that can support increasingly complex (and expensive) AI models has led engineers to explore new methods for expanding the computational capabilities of their machines.

Now Manya Ghobadi, an associate professor at MIT’s Department of Electrical Engineering and Computer Science (EECS) and a CSAIL member, and her colleagues have developed a solution: a system (dubbed “Lightning”) that connects photons (produced by lasers) to the electronic components (transistors and wires) of computers—creating hybrid photonic-electronic, reconfigurable network interface cards (“SmartNICs).

First photonic-computing prototype to serve real-time, machine-learning inference requests

The new system allows deep neural networks (machine-learning models that imitate how brains process information) to complete inference tasks (like image recognition and large language models generated in chatbots such as ChatGPT).

Machine-learning services completing inference-based tasks, like ChatGPT and BERT, currently require heavy computing resources and are expensive—some estimates show that ChatGPT requires $3 million or more per month to run.

They’re also environmentally detrimental. Instead, Lightning uses photons, which move faster than electrons do in wires while generating less heat.

The team is currently (Sept. 10–14) presenting their findings at the Association for Computing Machinery’s Special Interest Group on Data Communication (SIGCOMM).

Let us know your thoughts! Sign up for a Mindplex account now, join our Telegram, or follow us on Twitter.

How ecology and AI could collaborate

A paper published September 11 in the open-access journal Proceedings of the National Academy of Sciences argues for a synergy between AI and ecology—one that could strengthen AI and also help to solve complex global challenges, such as disease outbreaks, loss of biodiversity, and climate-change impacts.

Dealing with complex systems

The paper argues that there are many more possibilities for applying AI in ecology, such as in “synthesizing big data and finding missing links in complex systems,” said co-author Shannon LaDeau, a disease ecologist at Cary Institute.

Ecologists are also using AI to searching for patterns in large data sets and making more accurate predictions, such as whether new viruses might be capable of infecting humans and which animals are most likely to harbor those viruses.

Smarter AI

Inspired by ecological systems, a more robust AI might include feedback loops, redundant pathways, and decision-making frameworks, the researchers suggest. These flexibility upgrades could also contribute to a more “general intelligence” for AIs that could enable reasoning and connection-making beyond the specific data that the algorithm was trained on.

Ecology could also help the AI-driven large language models that power chatbots block “hallucinations” (when an AI generates false information). Because ecology examines complex systems at multiple levels and in holistic ways, it’s good at capturing emergent properties and can help reveal the mechanisms behind such behaviors, the researchers suggest.

The research was funded by the National Science Foundation.

Citation: Barbara A. Han et al. September 11, 2023. A synergistic future for AI and ecology. https://www.pnas.org/doi/10.1073/pnas.2220283120 (open source)

https://doi.org/10.1073/pnas.2220283120

Let us know your thoughts! Sign up for a Mindplex account now, join our Telegram, or follow us on Twitter.

Was this written by a human or AI? You probably can’t tell.

Even linguistics experts are largely unable to spot the difference between writing created by an AI or human, say University of South Florida researchers.

Research just published in the ScienceDirect journal Research Methods in Applied Linguistics revealed that experts from the world’s top linguistic journals could not differentiate between AI- and human-generated abstracts more than 39 percent of the time.

“We thought if anybody is going to be able to identify human-produced writing, it should be people in linguistics who’ve spent their careers studying patterns in language and other aspects of human communication,” said Matthew Kessler, a scholar in the USF the Department of World Languages, working with Elliott Casal, assistant professor of applied linguistics at The University of Memphis.

Linquists mostly clueless

So Kessler and associates tasked 72 experts in linguistics with reviewing a variety of research abstracts to determine whether they were written by AI or humans. Each expert was asked to examine four writing samples. None correctly identified all four; 13 percent got them all wrong.

Based on this, Kessler and Casal concluded ChatGPT can write short genres just as well as most humans—if not better in some cases: AI typically does not make grammatical errors.

Longer texts: probably generated by humans

But for longer texts, AI has been known to hallucinate and make up content, making it easier to identify that it was generated by AI,” Kessler said.

Kessler hopes this study will lead to a bigger conversation to establish the necessary ethics and guidelines surrounding the use of AI in research and education.

Citation: Casal, J. E., & Kessler, M. (2023). Can linguists distinguish between ChatGPT/AI and human writing?: A study of research ethics and academic publishing. Research Methods in Applied Linguistics, 2(3), 100068. https://doi.org/10.1016/j.rmal.2023.100068 (open-access)

Let us know your thoughts! Sign up for a Mindplex account now, join our Telegram, or follow us on Twitter.

Global cooperation urgently needed to govern risks of advanced AI, warns report

A new report by The Millennium Project warns that advanced artificial intelligence systems could emerge sooner than expected, posing unprecedented risks unless prudent governance frameworks are rapidly put in place.

The report, focused on International governance Issues of the transition from Artificial Narrow Intelligence to Artificial General Intelligence (AGI), is a distillation of interviews and collected insights from 55 global AI experts on 22 questions.

The experts on how to regulate AGI—AI that can handle novel situations as well as, or better than humans—say it represents an “intelligence explosion” that creates AI that surpasses human abilities, the report states.

New kind of flexible governance

“Lack of governance could lead to catastrophic outcomes, including existential threats to humanity if such systems are misaligned with human values and interests,” the report says. And “no existing governance models are adequately prepared to manage the risks and opportunities posed by artificial general intelligence (AGI).”

The report calls for rapid development of a new kind of flexible governance that can match and anticipate the pace of AI change and can provide the necessary safeguards while not stifling the promises and advance of AI.

“AGI is closer than any time before—the next advance could surpass human intelligence,” says Dr. Ilya Sutskever, co-founder of OpenAI, in the report.

“This might even happen within just a few years,” Dr. Ben Goertzel, author of AGI Revolution added. “It is more about who controls the development and use of AGI than a list of ethics.

“Alignment with human values is critical but challenging. I think we can create AGI by assembling components that already exist,” Goertzel suggested. “For example, connecting a large language model, a symbolic reasoning system, and an evolutionary learning system with the right combination architecture (such as our OpenCog Hyperon system running on SingularityNET).

“In this way, I believe, we can get an AGI with roughly human-level capability. This might even happen just a few years from now. If so, It would be capable of rewriting and improving itself—which may then lead to super-human-level AGI.”

Racing for AI supremacy

Because the benefits of AGI are so great in medicine, education, management, and productivity, corporations are racing to be first.

AGI will increase political power, so governments are also racing to be first.

International cooperation is essential but threatened by competitive tensions among nations and corporations.

The shared risks may compel collaboration, but overcoming distrust poses enormous challenges:

Extraordinary enforcement powers may be needed for governance to be trusted and effective globally, potentially including military capabilities.

Although controversial, proposals to limit research and development may be needed to allow time to design and implement management solutions.

The window for developing effective governance is short, demanding unprecedented collaboration.

Urgent action

The Millennium Project is calling for urgent action to create AGI governance and alignment at national and international levels before advanced AI exceeds humanity’s ability to control it safely. With stakes potentially including human extinction, the report warns we can ill afford delay in mobilizing global cooperation.

“If we don’t get an UN Convention on AGI and a UN AGI Agency to enforce rules, guardrails, auditing, and verification right, various forms of artificial superintelligence could emerge beyond our control and not to our liking,” says Jerome Glenn, CEO of The Millennium Project. “Management of AGI could be the most difficult problem humanity has ever faced.”

This work was supported by the Dubai Future Foundation and the Future of Life Institute. The Millennium Project is an international participatory think tank with 70 global nodes and three regional networks. Established in 1996, it has published more than 60 futures research projects based on international judgments.

Full disclosure: I participated with the Millennium Project in this study—Amara Angelica

Let us know your thoughts! Sign up for a Mindplex account now, join our Telegram, or follow us on Twitter.

Miniature eel-inspired bio-integrated devices capable of directly stimulating nerve cells for targeted drug therapies and speeding up wound healing

Researchers from the University of Oxford’s Department of Chemistry have made a significant step towards developing miniature powered bio-integrated devices that can interact with and stimulate cells, published today, Aug. 30, in an open-access paper in the journal Nature.

The researchers have developed a miniature power source capable of altering the activity of cultured human nerve cells. Inspired by how electric eels generate electricity, the device uses internal ion gradients to generate energy, rather than electrons.

Important therapeutic applications include delivering targeted drug therapies and speeding up wound healing.

How the technology works

The miniaturized soft-power source is produced by depositing a chain of five nanoliter-sized droplets of a conductive hydrogel (a 3D network of polymer chains containing a large quantity of absorbed water.

Each droplet has a different composition so that a salt concentration gradient is created across the chain. The droplets are separated from their neighbors by lipid bilayers, which provide mechanical support while preventing ions from flowing between the droplets.

The power source is turned on by cooling the structure to 4°C and changing the surrounding medium: this disrupts the lipid bilayers and causes the droplets to form a continuous hydrogel.

This allows the ions to move through the conductive hydrogel, from the high-salt droplets at the two ends to the low-salt droplet in the middle. By connecting the end droplets to electrodes, the energy released from the ion gradients is transformed into electricity, enabling the hydrogel structure to act as a power source for external components.

In the study, the research team demonstrated how living cells could be attached to one end of the device so that their activity could be directly regulated by the ionic current. The team attached the device to droplets containing human neural progenitor cells that had been stained with a fluorescent dye to indicate their activity. When the power source was turned on, time-lapse recording demonstrated waves of intercellular calcium signaling* in the neurons, induced by the local ionic current.

“The miniaturized soft power source represents a breakthrough in bio-integrated devices, Dr. Yujia Zhang (Department of Chemistry, University of Oxford), the lead researcher for the study, said. By harnessing ion gradients, we have developed a miniature, biocompatible system for regulating cells and tissues on the microscale, which opens up a wide range of potential applications in biology and medicine.”

Next-generation wearable devices, bio-hybrid interfaces, implants, synthetic tissues, and microrobots

According to the researchers, the device’s modular design would allow multiple units to be combined to increase the voltage and/or current generated. This could open the door to powering next-generation wearable devices, bio-hybrid interfaces, implants, synthetic tissues, and microrobots.

By combining 20 five-droplet units in series, they were able to illuminate a light-emitting diode, which requires about 2 Volts. They envisage that automating the production of the devices by using a droplet printer, for instance, could produce droplet networks composed of thousands of power units.

“This work addresses the important question of how stimulation produced by soft, biocompatible devices can be coupled with living cells. The potential impact on devices, including bio-hybrid interfaces, implants, and microrobots is substantial,” said Professor Hagan Bayley, University of Oxford.

* Calcium signaling is a key mechanism through which neurons communicate to one another to coordinate biological activities such as neurotransmitter release, neuronal firing, synaptic plasticity, and gene transcription.

Citation: Zhang, Y., Riexinger, J., Yang, X., Mikhailova, E., Jin, Y., Zhou, L., & Bayley, H. (2023). A microscale soft ionic power source modulates neuronal network activity. Nature, 620(7976), 1001-1006. https://doi.org/10.1038/s41586-023-06295-y

Let us know your thoughts! Sign up for a Mindplex account now, join our Telegram, or follow us on Twitter.

Yes, metal really can heal itself. Seriously. (An update)

In a recent experiment in July at Sandia National Laboratories, a microscopic crack grew in a very small piece of platinum when placed under repetitive stretching. The experiment, designed to study fatigue crack growth, continued as predicted for a while, before something unexpected happened.

The crack apparently stopped growing. Instead, it began to get shorter, effectively “healing” itself, researchers at Sandia National Laboratories claimed. We (Mindplex) took their word for it. We published a story on their finding, mainly because it was reported in the journal Nature.

Dr. Michael Demkowicz, a professor in Texas A&M University’s materials science and engineering department and coauthor of the recent study, explained that nanocrystalline metals make studying self-healing easier because their small grain size allows for more microstructural features that even small cracks can interact with.

Spacefaring technology

He now suggests that self-healing could be possible in conventional metals with larger grain sizes, but future investigations will be needed.

His studies (at Sandia National Laboratories and previously at MIT 10 years ago) found that one such feature, grain boundaries, can affect crack healing depending on the direction of boundary migration relative to the crack. Demkowicz explained that these features are common in many metals and alloys and can be manipulated.

But one condition common to both the 2013 theory and the recent experiment: both were conducted in vacuum environments, devoid of foreign matter, explain the researchers. Such outside matter could interfere with crack surfaces’ ability to bond, or cold-weld, back together.

Even with this limitation, applications could still be possible for spacefaring technology or internal cracks that are not exposed to outside air.

Citation: Barr, C. M., Duong, T., Bufford, D. C., Milne, Z., Molkeri, A., Heckman, N. M., Adams, D. P., Srivastava, A., Hattar, K., Demkowicz, M. J., & Boyce, B. L. (2023). Autonomous healing of fatigue cracks via cold welding. Nature, 1-5. https://doi.org/10.1038/s41586-023-06223-0

Let us know your thoughts! Sign up for a Mindplex account now, join our Telegram, or follow us on Twitter.

Ultra-thin battery charged by saline solution could power smart augmented-reality contact lenses

Imagine lenses that flash data on our corneas, keep an eye on our health, or flag diseases like glaucoma or diabetes

A new flexible battery as thin as a human eye’s cornea could power future smart-contact lenses, which in the future will be capable of displaying visible information on a eye’s cornea. It could be used to access augmented-reality devices, such as the Apple Vision Pro, for example.

Developed by researchers at Nanyang Technological University (NTU) in Singapore, the battery can be recharged simply by immersing it in saline solution. (Existing rechargeable batteries rely on wires or induction coils that contain metal so they are uncomfortable and present risks to the human eye.)

How they work

The new battery, made of biocompatible materials, does not contain wires or toxic heavy metals, such as those in lithium-ion batteries or wireless charging systems.

It has a glucose-based coating that reacts with the sodium and chloride ions in the saline solution surrounding it (water contained in the battery serves as the “wire” or “circuitry” for electricity to be generated).

Powered by tears

“This research began with a simple question: could contact lens batteries be recharged with our tears?” said Associate Professor Lee Seok Woo, from NTU’s School of Electrical and Electronic Engineering (EEE), who led the study. “Previous techniques for lens batteries were not perfect as one side of the battery electrode was charged and the other was not. The battery could be powered by human tears because they contain sodium and potassium ions, at a low concentration.

Testing the current battery with a simulated tear solution, the researchers showed that the battery’s life would be extended one additional hour for every 12-hour wearing cycle it is used. The battery can even be charged conventionally by an external power supply.

Takes just glucose and water to generate electricity

“Our approach can charge both electrodes of a battery through a unique combination of enzymatic reaction and self-reduction reaction. It also relies on just glucose and water to generate electricity, both of which are safe to humans and would be less harmful to the environment when disposed, compared to conventional batteries. The battery, which is about 0.5 millimetrers-thin, generates electrical power by reacting with the constant tears that create a thin film over our eyeballs.”

The NTU team will be conducting further research to improve the amount of electrical current their battery can discharge. They will also be working with several contact lenses companies to implement their technology.

Citation: Yun, J., Li, Z., Miao, X., Li, X., Lee, J. Y., Zhao, W., & Lee, S. W. (2023). Jeonghun Yun et al. A tear-based battery charged by biofuel for smart contact lenses, Nano Energy, Jun 2023. DOI 1016/j.nanoen.2023.108344 (open-access)

Let us know your thoughts! Sign up for a Mindplex account now, join our Telegram, or follow us on Twitter.

Meta has created an AI model, ‘SeamlessM4T,’ that can translate and transcribe close to 100 languages across text and speech

The model is “the first all-in-one multilingual multimodal AI translation and transcription model,” Meta claims.

“It can perform speech-to-text, speech-to-speech, text-to-speech, and text-to-text translations for up to 100 languages, depending on the task … without having to first convert to text behind the scenes, among other. We’re developing AI to eliminate language barriers in the physical world and in the metaverse.”

“In keeping with our approach to open science, we’re publicly releasing SeamlessM4T under a research license to allow researchers and developers to build on this work. We’re also releasing the metadata of SeamlessAlign, the biggest open multimodal translation dataset to date, totaling 270,000 hours of mined speech and text alignments.”

Competition

“Beyond the wealth of commercial services and open source models already available from Amazon, Microsoft, OpenAI and a number of startups, Google is creating what it calls the Universal Speech Model, a part of the tech giant’s larger effort to build a model that can understand the world’s 1,000 most-spoken languages,” says TechCrunch.

“Tech Mozilla meanwhile spearheaded Common Voice, one of the largest multi-language collections of voices for training automatic speech recognition algorithms. But SeamlessM4T is among the more ambitious efforts to date to combine translation and transcription capabilities into a single model.

In developing it, Meta says that it scraped publicly available text (on the order of ‘tens of billions’ of sentences) and speech (4 million hours) from the web.”

Let us know your thoughts! Sign up for a Mindplex account now, join our Telegram, or follow us on Twitter.

Exciting News! Our Mobile App is Here!

Welcome Back

No account? Create One

Join

Already have an account? Sign in

forgot password