Artificial intelligence
fromThe Atlantic
1 day agoThe AI Industry Wants to Automate Itself
Protesters in San Francisco demand a halt to the development of self-improving AI technologies, fearing existential risks to humanity.
Generative AI is now incorporated into the workflow for many scholars across many disciplines, but the broader scientific community would benefit from taking stock of how this technology could truly benefit our work and how it might distract. We hope the symposium can provide clarity.
Time pressure, limited information, confusion, fatigue, and mortality salience combine to set the stage for decision-making errors, sometimes with grave consequences. An example is the downing of Iran Air Flight 655 by a missile launched by the USS Vincennes in 1988, resulting in the death of 290 passengers and crew. In a time of heightened tension between the U.S. and Iran, the captain of the Vincennes misidentified the airliner as an incoming hostile aircraft and ordered his crew to shoot it down.
According , when ICE identifies a recruit with prior law enforcement experience, it assigns them to its "Law Enforcement Officer Program." This is a four-week online course meant to streamline training for those already familiar with the legal aspects of the gig. Everyone else gets shipped off to ICE's Federal Law Enforcement Training Center in Georgia for an eight-week in-person academy. This more rigorous training includes courses in immigration law, gun handling, physical fitness exams, and more.
Biology is undergoing a transformation. After centuries of studying life as it evolves naturally, researchers are now using a combination of computation and genome engineering to intervene, generating new proteins and even whole bacteria from scratch. The use of artificial-intelligence tools to design biological components, an approach known as generative biology, is set to turbocharge this area of research. Just last year, scientists used AI-assisted design to produce artificial genes that can be expressed in mammalian cells.
Which Algorithm Is This? If you step back, this maps almost perfectly to the Top K Frequent Elements problem.We usually solve it for integers in a list. Here, the "elements" are audience profiles age and body-type combinations. First, define what an audience profile looks like: case class Profile(age: Int, height: Int, weight: Int) What we want is a function like this:
This is a state where we see that the teams that move fastest will be the ones with clear tests, tight review policies, automated enforcement and reliable merge paths. Those guardrails are what make AI useful. If your systems can automatically catch mistakes, enforce standards, and prove what changed and why, then you can safely let agents do the heavy lifting. If not, you're just accelerating risk,
A dyad has three parts, not two: Partner A, Partner B, and the relationship or agreements between them. A dyad of two experts who cannot communicate clearly will often lose to a dyad of less-skilled individuals who coordinate effectively.
Called AlphaGenome, the model could help scientists discover why subtle differences in our DNA put us at risk of conditions such as high blood pressure, dementia and obesity. It could also dramatically accelerate our understanding of genetic diseases and cancer. The developers of the model acknowledge it's not perfect, but experts have described it as "an incredible feat" and "a major milestone".
The scaling model relies on several predictive factors of the system, including the underlying LLM's intelligence index; the baseline performance of a single agent; the number of agents; number of tools; and coordination metrics. The researchers found there were three dominant effects in the model: tool-coordination trade-off, where tasks requiring many tools perform worse with multi-agent overhead; capability saturation, where adding agents yields diminishing returns when the single-agent baseline performance exceeds a certain threshold; and topology-dependent error amplification, where centralized orchestration reduces error amplification.
Fifty-four seconds. That's how long it took Raphael Wimmer to write up an experiment that he did not actually perform, using a new artificial-intelligence tool called Prism, released by OpenAI last month. "Writing a paper has never been easier. Clogging the scientific publishing pipeline has never been easier," wrote Wimmer, a researcher in human-computer action at the University of Regensburg in Germany, on Bluesky. Large language models (LLMs) can suggest hypotheses, write code and draft papers, and AI agents are automating parts of the research process.
When a scientist feeds a data set into a bot and says "give me hypotheses to test", they are asking the bot to be the creator, not a creative partner. Humans tend to defer to ideas produced by bots, assuming that the bot's knowledge exceeds their own. And, when they do, they end up exploring fewer avenues for possible solutions to their problem.