DeepMind AI integrates fact checker to make groundbreaking mathematical findings

DeepMind’s FunSearch AI can tackle mathematical problems

Arengo/Getty Images

Google DeepMind claims to have made the first ever scientific discovery in an AI chatbot by building a fact checker that filters out useless output and leaves behind only reliable solutions to mathematical or computing problems. Masu.

DeepMind’s previous achievements, such as using AI to predict the weather or the shape of proteins, rely on models created specifically for the task at hand and trained on accurate, specific data. I did. Large-scale language models (LLMs), such as GPT-4 and Google’s Gemini, are instead trained on vast amounts of disparate data, yielding a wide range of capabilities. However, this approach is also susceptible to “hallucinations,” which refers to researchers producing erroneous output.

Gemini, released earlier this month, has already shown hallucination tendencies and even gained simple facts such as: This year’s Oscar winners were wrong. Google’s previous AI-powered search engine even had errors in its self-launched advertising materials.

One common fix for this phenomenon is to add a layer on top of the AI ​​that validates the accuracy of the output before passing it on to the user. However, given the wide range of topics that chatbots may be asked about, creating a comprehensive safety net is a very difficult task.

Al-Hussein Fawzi Google’s DeepMind and his colleagues created a general-purpose LLM called FunSearch based on Google’s PaLM2 model with a fact-checking layer they call an “evaluator.” Although this model is constrained by providing computer code that solves problems in mathematics and computer science, DeepMind says this work is important because these new ideas and solutions are inherently quickly verifiable. is a much more manageable task.

The underlying AI may still hallucinate and provide inaccurate or misleading results, but the evaluator filters out erroneous outputs, leaving only reliable and potentially useful concepts. .

“We believe that probably 90% of what LLM outputs is useless,” Fawzi says. “If you have a potential solution, it’s very easy to tell whether this is actually the correct solution and evaluate that solution, but it’s very difficult to actually come up with a solution. So , mathematics and computer science are a particularly good fit.”

DeepMind claims the model can generate new scientific knowledge and ideas, something no LLM has ever done before.

First, FunSearch is given a problem and a very basic solution in the source code as input, and then generates a database of new solutions that are checked for accuracy by evaluators. The best reliable solutions are returned as input to the LLM with prompts to improve the idea. According to DeepMind, the system generates millions of potential solutions and eventually converges on an efficient result, sometimes even exceeding the best known solution.

For mathematical problems, a model creates a computer program that can find a solution, rather than trying to solve the problem directly.

Fawzi and his colleagues challenged FunSearch to find a solution to the cap set problem. This involves determining the pattern of points where three points do not form a straight line. As the number of points increases, the computational complexity of the problem increases rapidly. The AI ​​discovered a solution consisting of 512 points in eight dimensions, larger than previously known.

When tackling the problem of bin packing, where the goal is to efficiently place objects of different sizes into containers, FunSearch discovered a solution that outperformed commonly used algorithms. The result is a result that can be immediately applied to transportation and logistics companies. DeepMind says FunSearch could lead to improvements in more math and computing problems.

mark lee The next breakthrough in AI will not be in scaling up LLM to ever-larger sizes, but in adding a layer to ensure accuracy, as DeepMind has done with FunSearch, say researchers at the University of Birmingham, UK. It is said that it will come from.

“The strength of language models is their ability to imagine things, but the problem is their illusions,” Lee says. “And this study breaks that down, curbs that, and confirms the facts. It’s a nice idea.”

Lee says AI should not be criticized for producing large amounts of inaccurate or useless output. This is similar to how human mathematicians and scientists work: brainstorm ideas, test them, and follow up on the best while discarding the worst.

topic:

Source: www.newscientist.com

Thermal secrets uncovered in neutron star mergers through gravitational waves

by

Scientists used supercomputer simulations to study gravitational waves produced by neutron star mergers and found a correlation between residual temperature and gravitational wave frequency. These findings are important for future gravitational wave detectors that distinguish models of hot nuclear material. Credit: SciTechDaily.com

Binary simulation neutron star This merger suggests that future detectors will distinguish between different models of hot nuclear material.

Researchers used supercomputer simulations to investigate the effects of neutron star mergers gravitational waves, found a significant relationship with debris temperature. This research will aid future advances in the detection and understanding of hot nuclear materials.

Exploring neutron star mergers and gravitational waves

When two neutron stars orbit each other, they emit ripples into spacetime called gravitational waves. These ripples drain energy from the orbit until the two stars eventually collide and combine into one object. Scientists used supercomputer simulations to investigate how the behavior of different models of nuclear material affects the gravitational waves released after these mergers. They found a strong correlation between the temperature of the debris and the frequency of these gravitational waves. Next generation detectors will be able to distinguish these models from each other.

Plot comparing density (right) and temperature (left) for two different simulations (top and bottom) of a neutron star merger, viewed from above, approximately 5 ms after the merger.Credit: Jacob Fields, Pennsylvania State University

Neutron Star: Institute for Nuclear Materials

Scientists use neutron stars as laboratories for nuclear materials under conditions that would be impossible to explore on Earth. They will use current gravitational wave detectors to observe neutron star mergers and learn how cold, ultra-dense matter behaves. However, these detectors cannot measure the signal after the stars have merged. This signal contains information about hot nuclear material. Future detectors will be even more sensitive to these signals. Because different models can also be distinguished from each other, the findings suggest that future detectors could help scientists create better models of hot nuclear material.

Detailed analysis of neutron star mergers

The study investigated neutron star mergers using THC_M1, a computer code that simulates neutron star mergers and accounts for the bending of spacetime due to the star’s strong gravitational field and neutrino processes in dense matter. . The researchers tested the effect of heat on mergers by varying the specific heat capacity of the equation of state, which measures the amount of energy required to raise the temperature of neutron star material by one degree Celsius. To ensure the robustness of their results, the researchers ran their simulations at two resolutions. They repeated the high-resolution run using a more approximate neutrino processing.

References:

“Thermal effects in binary neutron star mergers” by Jacob Fields, Aviral Prakash, Matteo Breschi, David Radice, Sebastiano Bernuzzi, and Andre da Silva Schneider, July 31, 2023. of Astrophysics Journal Letter.
DOI: 10.3847/2041-8213/ace5b2

“Identification of nuclear effects in neutrino-carbon interactions in low 3 momentum transfer” until February 17, 2016 physical review letter.
DOI: 10.1103/PhysRevLett.116.071802

Funding: This research was primarily funded by the Department of Energy, Office of Science, Nuclear Physics Program. Additional funding was provided by the National Science Foundation and the European Union.

This research used computational resources available through the National Energy Research Scientific Computing Center, the Pittsburgh Supercomputing Center, and the Pennsylvania State University Computing and Data Science Institute.

Source: scitechdaily.com

Revolutionary New Technology Accelerates Diaper Recycling by 200 Times

The superabsorber becomes a liquid under ultraviolet light after absorbing enough water. It can then be reused. Credit: Ken Pekarsky, KIT

Water and UV light effectively and quickly break down the cross-linked polymers in diaper liners without the use of any chemicals. This process allows recycled plastic molecules to be reused for a variety of uses.

Superabsorbent materials such as sodium polyacrylate are important components of a variety of hygiene and medical products, including diapers, bandages, and dressings. These cross-linked polymers are typically insoluble in water, although they are known for their high absorbency. Recycling them traditionally required the use of strong acids.

It will not melt at high temperatures, it will only deteriorate. However, the acid “breaks” the chains and stabilizes the polymer after about 16 hours at 80 degrees. Celsius Therefore, recycling is now possible. Because this process is complex and expensive, superabsorbents are rarely recycled. Approximately 2 million tons of this waste is thrown away or incinerated each year.

Turns into liquid in 5 minutes instead of 16 hours

Researchers from KIT’s Institute of Biochemical Systems, Institute of Biointerfaces, and Institute of Chemical Technology and Polymer Chemistry have discovered that crosslinked sodium polyacrylate polymers degrade under ultraviolet light after uptake of water. .

“The chains that connect the polymers are broken by light, and they are so loose that they swim underwater and turn into liquid fibers,” explains Pavel Levkin, a professor at the Institute of Biochemical Systems. For the study, researchers cut liners from traditional diapers, wetted them with water and exposed them to a 1000 W lamp. After 5 minutes, the solid material turned into a liquid and fell into the collector. “This method using ultraviolet light is about 200 times faster than using acids,” Revkin says.

Recycled polymers can be used in a variety of ways

The team then used known processes to convert the liquid into new adhesives and dyes. “The observation that this substance is soluble and processable was very important. It could probably be turned into many other products,” explains the scientist.

In the test, the researchers used clean diapers. However, it is also possible to separate superabsorbents from used diapers. “Therefore, there is no reason why a near-realistic use should not be possible,” Revkin says. By using solar power, you can optimize recycling methods that are cost-effective and environmentally friendly. “We have discovered a promising strategy to recycle superabsorbents, which significantly reduces environmental pollution and contributes to a more sustainable use of polymers.”

Reference: “From diapers to thickeners and pressure-sensitive adhesives: recycling superabsorbents by UV degradation” by Shuai Li, Johannes M. Scheiger, Zhenwu Wang, Birgit Huber, Maxi Hoffmann, Manfred Wilhelm, Pavel A. Levkin , September 7, 2023 ACS Applied Materials & Interfaces.
DOI: 10.1021/acsami.3c06999

Source: scitechdaily.com