Google DeepMind's new AI system can now solve complex math problems
Rhiannon Williams | MIT Technology Review
“AI models can easily generate essays and other types of text. But they don't come close to solving math problems that require logical reasoning — something that's beyond the capabilities of most current AI systems. But that may finally be changing: Google DeepMind says it has trained two specialized AI systems to solve complex math problems that require advanced reasoning.”
This startup is building the nation's most powerful quantum computer on Chicago's south side
Adam Bluestein | Fast Company
“PsiQuantum's approach is fundamentally different from that of its competitors: it uses cutting-edge 'silicon photonics' to manipulate single particles of light and perform calculations. And rather than taking a piecemeal approach to building a supercomputer, the company is entirely focused on starting with a full-scale 'fault-tolerant' system that will be far larger than any quantum computer built to date. The company has pledged to have its first system operational by late 2027, years earlier than other predictions.”
The race to the next Ozempic
Emily Marin | Wired
“These drugs are now extremely popular, resulting in shortages and huge profits for their manufacturers. The success of these drugs has created a frenzy among pharmaceutical companies searching for the next blockbuster weight-loss drug. Researchers are now racing to develop new anti-obesity drugs that are more effective, more convenient and have fewer side effects than those currently on the market.”
Watch as a robot peels a pumpkin with human-like dexterity
Alex Wilkins | New Scientist
“Pulkit Agrawal and his colleagues at the Massachusetts Institute of Technology have developed a robotic system that can rotate different kinds of fruits and vegetables using the fingers of one hand and peel them with the other.”
What happens when you give people free money?
Paresh Dave | Wired
“Early findings from what Open Research, a research institute funded by Altman, calls the most comprehensive study of 'unconditional cash', suggest that while the grants have benefits and aren't used for things like drugs or alcohol, they're not a panacea for addressing some of the biggest concerns about income inequality and the potential for AI and other automation technologies to take over jobs.”
Meta Releases Its Largest and Best Open Source AI Model Yet
Alex Heath | The Verge
“Meta releases Llama 3.1, the largest open-source AI model ever made. The company claims the model outperforms GPT-4o and Anthropic's Claude 3.5 Sonnet on several benchmarks. … CEO Mark Zuckerberg predicts Meta AI will surpass ChatGPT by the end of the year to become the most widely used assistant.”
US solar power generation up 25% in just one year
John Timmer | Ars Technica
“In terms of utility-scale production, the first five months of 2024 were up 29% compared to the same period last year. Small-scale solar was up 'only' 18%, for a total increase of 25.3%. … It's worth noting that all of this data comes before the most productive month of the year for solar. Overall, the EIA projects that solar production could increase by up to 42% in 2024.”
SearchGPT is a direct attack on Google by OpenAI
Reese Rogers and Will Knight | Wired
“After months of speculation about its search ambitions, OpenAI has announced SearchGPT, a 'prototype' search engine that could finally help the company steal some of Google's lucrative business. OpenAI says the new tool uses generative AI to collect links and answer users' questions in a conversational tone to help them find what they're looking for faster and easier.”
Thin light sails could help us reach another star faster
Alex Wilkins | New Scientist
“The light sail, designed with artificial intelligence, is about 1/1000th the thickness of a human hair and weighs about the same as a grain of sand. And this could help us build spacecraft that can reach another star faster than we thought possible.”
art
AI can't make music
Mateo Wong | The Atlantic
“While AI models are beginning to replicate musical patterns, it's the breaking of the rules that tends to produce era-defining songs. Algorithms are 'good at meeting expectations, but not at subverting them,' and yet that often produces the best music,” Eric Drott, a professor of music theory at the University of Texas at Austin, told me.
Image credit: David Clode / Unsplash
