Lifeboat Foundation: Safeguarding Humanity
Toggle light / dark theme

Blog

Category: robotics/AI – Page 593

We’re putting AI brains in robot bodies now. What could go wrong?

Brooks himself is among the philosophers who have previously said giving AI sensory and motor skills to engage with the world may be the only way to create true artificial intelligence. A good deal of human creativity, after all, comes from physical self-preservation — a caveman need only cut himself once on sharpened bone to see its use in hunting. And what is art if not a hope that our body-informed memories may outlive the body with which we formed them?

If you want to get even more mind-bent, consider thinkers like Lars Ludwig, who proposed that memory isn’t even something we can hold exclusively in our bodies anyway. Rather, to be human always meant sharing consciousness with technology to “extend artificial memory” — from a handprint on a cave wall, to the hard drive in your laptop. Thus, human cognition and memory could be considered to take place not just in the human brain, nor just in human bodily instinct, but also in the physical environment itself.

India reverses AI stance, requires government approval for model launches

India has waded into global AI debate by issuing an advisory that requires “significant” tech firms to get government permission before launching new models.

India’s Ministry of Electronics and IT issued the advisory to firms on Friday. The advisory — not published on public domain but a copy of which TechCrunch has reviewed — also asks tech firms to ensure that their services or products “do not permit any bias or discrimination or threaten the integrity of the electoral process.”

Though the ministry admits the advisory is not legally binding, India’s IT Deputy Minister Rajeev Chandrasekhar says the notice is “signalling that this is the future of regulation.” He adds: “We are doing it as an advisory today asking you to comply with it.”

Harder To Outrun China’s New Humanoid Robot

Unitree is a publicly traded robot company with about $5 billion in market value. They have sped up their humanoid robot to human jogging speed of 3.3 meters per second. This is about 7.5 miles per hour. It would not tire so it would take 50 minutes to cover a 10 kilometer race with enough battery power.

It can lift boxes and climb and descend stairs. It was able to jump vertically.

They have hand attachments that currently do not have finger and grasping motions.

Researchers Create AI-Powered Malware That Spreads on Its Own

Researchers have developed a computer “worm” that can spread from one computer to another using generative AI, a warning sign that the tech could be used to develop dangerous malware in the near future — if it hasn’t already.

As Wired reports, the worm can attack AI-powered email assistants to obtain sensitive data from emails and blast out spam messages that infect other systems.

“It basically means that now you have the ability to conduct or to perform a new kind of cyberattack that hasn’t been seen before,” Cornell Tech researcher Ben Nassi, coauthor of a yet-to-be-peer-reviewed paper about the work, told Wired.

AI vs. Cancer: A Game-Changer!

MIT and Dana-Farber Cancer Institute have teamed up to create an AI model that CRACKS the code of mysterious cancer origins! No more guesswork-this model predicts where tumors come from with up to 95% accuracy. For more insight, visit https://www.channelchek.com #Cancer #CancerBreakthrough #AIinMedicine #MedicalScience #BioTech #FutureOfHealthcare #FightCancer #HealthTech #CancerResearch #PrecisionMedicine

New research shows how child-like language learning is possible using AI tools

AI systems, such as GPT-4, can now learn and use human language, but they learn from astronomical amounts of language input—much more than children receive when learning how to understand and speak a language. The best AI systems train on text with a word count in the trillions, whereas children receive just millions per year.

Due to this enormous data gap, researchers have been skeptical that recent AI advances can tell us much about human learning and development. An ideal test for demonstrating a connection would involve training an AI model, not on massive data from the web, but on only the input that a single child receives. What would the model be able to learn then?

A team of New York University researchers ran this exact experiment. They trained a multimodal AI system through the eyes and ears of a single child, using headcam video recordings from when the child was 6 months and through their second birthday. They examined if the AI model could learn words and concepts present in a child’s everyday experience.

Using large language models to accurately analyze doctors’ notes

The amount of digital data available is greater than ever before, including in health care, where doctors’ notes are routinely entered into electronic health record systems. Manually reviewing, analyzing, and sorting all these notes requires a vast amount of time and effort, which is exactly why computer scientists have developed artificial intelligence and machine learning techniques to infer medical conditions, demographic traits, and other key information from this written text.

However, safety concerns limit the deployment of such models in practice. One key challenge is that the medical notes used to train and validate these models may differ greatly across hospitals, providers, and time. As a result, models trained at one hospital may not perform reliably when they’re deployed elsewhere.

Previous seminal works by Johns Hopkins University’s Suchi Saria—an associate professor of computer science at the Whiting School of Engineering—and researchers from other top institutions recognize these “dataset shifts” as a major concern in the safety of AI deployment.