The path towards generative artificial intelligence (AGI) requires a different approach to today’s generative AI models, inspired by natural ecosystems, argues a new company that has gone out of business mode. stealth and launched such a project.
Why this matters: Digital intelligence based on a network of intelligent agents is potentially cheaper, more environmentally sustainable, and more geopolitically defensible than a vast system trained on billions of data points.
Verses AI is entering the debate with the bold claim that the machine learning methods behind ChatGPT and the rest of the AI advancements of the last 20 years will never lead the industry to AGI.
Artificial intelligence learning like a human
AGI is the industry’s grail—a human-like level of artificial intelligence that can reason and learn in new ways—and Verses’ founders say today’s most advanced large language models, like OpenAI’s GPT-4, cannot deliver it.
“There is no evidence of ability to act outside of training data”CEO Gabriel René told Axios.
The big picture: Verses is working instead on what it calls distributed intelligence, using biology as a starting point, in the belief that AGI is only possible with a system that can self-organize and re-train itself in real time – like biological organisms do.
Verses chief scientist Karl Friston is betting that this requires higher degrees of autonomy and computational efficiency than the current school of large model development allows.
Genius – a more advanced learning system
Based on 30 scientific papers from its researchers, the company developed Genius, an operating system for “constantly learning autonomous agents” operating at the edge of our connected devices.
NASA’s Jet Propulsion Laboratory and Volvo are among Genius’ beta users.
“Minimizing complexity” is its “fundamental drive”, says Friston. Instead of building ever larger AI models, Verses aims to deliver “99% smaller models” without sacrificing quality.
“90% of the neural network is not useful. There must be a shift from big data to very well-curated sparse data.”Friston said, including the ability to forget data that is not relevant.
Flashback: AI’s roughly 70-year scientific journey has had many twists and turns, and just as the conventional wisdom of 25 years ago was upended by the emergence of machine learning, the dominant paradigm of massive scale today may be replaced by a different approach.
The history of computing is full of oscillations between monolithic and distributed systems, the center and the edge.
Jeff Hawkins, the inventor of the PalmPilot, was one of the pioneers of the theory of distributed intelligence – including a “thousand brains theory”.
Roboticist Rodney Brooks pioneered a distributed approach to autonomous machines that was featured in the Errol Morris documentary “Fast, Cheap and Out of Control.”
A game changer in artificial intelligence
René offered two analogies for the current moment in AI.
The contest between advocates of big AGI versus a network of smaller intelligent agents is “AOL versus the web” all over again, he says.
The world of AI, he predicts, is also heading towards a “smartphone moment” where AI equivalents from Nokia, Blackberry and Motorola are quickly replaced by better architecture.
The problem today is that “most of the AI out there is just good input-output mapping, with no sense of agency or context sensitivity” that real agents would display, Friston said.
Today’s AI models are pre-trained and memorize content, reproducing it in new formations as required – but because these systems lack critical thinking skills, they lack autonomy.
Only a model that can identify its errors and correct them by re-training in real time would qualify as “superintelligence,” René said.
How it works: Friston proposes an approach called “active inference,” in which intelligent agents with predictive and adaptive abilities autonomously share knowledge with other agents and generate new agents, creating a self-sustaining intelligence network.
Data x Beliefs
Friston argues that AGI cannot be achieved without belief structures. “I have to model my beliefs, not just the data”including quantifying uncertainty, just as humans try to assess what they don’t know when making judgments, he said.
“Instead of just scaling a machine, you are growing an ecosystem”René said, in which agents are gathering evidence for their own use and exchange with other agents – just like organisms in a human body do.
Ninety percent of your body is autonomous agents that work without your input, and your body is constantly seeking balance, René said.
Verses’ approach imagines a constant process of rebalancing between agents, similar to the balance achieved within Wikipedia through the interactions of the people who build it.
Point of friction: Verses has suggested that the market leader in generative AI, OpenAI, is violating its charter by not engaging with Verses.
Verses ran a full-page ad in the New York Times in December that announced its progress and methods. Verses challenged OpenAI to cooperate with it – asking the nonprofit company to fulfill its letter promise to “stop competing” with any “values-aligned, safety-conscious project that comes close to building AGI before we do.” ”.
Reality: Verses, based in Culver City, California, has about 100 employees in 60 remote locations and has raised $65 million.
It’s a small player entering a battlefield of giants, and even if its founders are right in theory, there’s no guarantee their approach will take off in the market.
What’s next: Verses says it will launch a public beta of Genius in the summer.
Source: Atrevida
Earl Johnson is a music writer at Gossipify, known for his in-depth analysis and unique perspective on the industry. A graduate of USC with a degree in Music, he brings years of experience and passion to his writing. He covers the latest releases and trends, always on the lookout for the next big thing in music.
