The researchers trained a trillion-parameter model using zero reinforcement learning to see how reasoning capabilities emerge at a massive scale without relying on human-annotated data. They found that pushing the parameter count to a trillion drastically improves both sample efficiency and the overall performance ceiling when compared to a smaller 104-billion parameter baseline strongly validating the concept that raw scale and computation eventually outpace hand-crafted human heuristics.

They also discovered that the training process reliably unfolds in two distinct sequential stages. The model starts with a discovery phase where it actively expands its reasoning boundaries by unlocking dormant pathways, and then it moves into a sharpening phase where it refines its policy within those established limits. Notably, the model spontaneously developed advanced cognitive strategies entirely on its own.

It began using structured formatting, parallel reasoning, self-verification, context anxiety, and even anthropomorphic expressions of frustration during complex tasks without any explicit human prompting. To keep the training stable at such a massive scale, the team relied on simple optimization techniques like clipped importance sampling and mixed-precision control. They also created a new evaluation framework to judge the actual quality of the reasoning steps based on comprehensibility, reproducibility, and token efficiency instead of just looking at the final answer.

From the who-could've-seen-this-coming dept:

Dozens of Meta employees have sued the social media company over claims that it used artificial intelligence tools to tag workers for mass layoffs. The workers allege that those AI tools targeted them after they asked for protected or maternity leave or disability accommodation.

The lawsuit, filed Monday in federal court in the northern district of California, points to Meta’s workforce reduction of about 8,000 employees earlier this year. Meta is the parent company of Facebook, Instagram and WhatsApp. The suit alleges that Meta used a “constellation of internal artificial intelligence systems”, including AI performance ratings and keystroke- and activity-monitoring data, to pinpoint who to lay off.

“Meta did not assemble the termination list through the considered judgment of managers who knew the work,” reads the 71-page complaint. Instead, the 26 workers listed in the lawsuit allege the company used AI systems “to score, rank and select employees for inclusion on the list”.

The plaintiffs are seeking a preliminary court ruling to stop Meta from finalizing the layoffs while they pursue their ​claims, along with relief that could include reinstatement, back pay, lost equity, benefits and other damages.

See also:

Before performing the study, the developers in question expected the AI tools would lead to a 24 percent reduction in the time needed for their assigned tasks. Even after completing those tasks, the developers believed that the AI tools had made them 20 percent faster, on average. In reality, though, the AI-aided tasks ended up being completed 19 percent slower than those completed without AI tools.

By analyzing screen recording data from a subset of the studied developers, the METR researchers found that AI tools tended to reduce the average time those developers spent actively coding, testing/debugging, or “reading/searching for information.” But those time savings were overwhelmed in the end by “time reviewing AI outputs, prompting AI systems, and waiting for AI generations,” as well as “idle/overhead time” where the screen recordings show no activity.

Overall, the developers in the study accepted less than 44 percent of the code generated by AI without modification. A majority of the developers reported needing to make changes to the code generated by their AI companion, and a total of 9 percent of the total task time in the “AI-assisted” portion of the study was taken up by this kind of review.

[...]

On the surface, METR’s results seem to contradict other benchmarks and experiments that demonstrate increases in coding efficiency when AI tools are used. But those often also measure productivity in terms of total lines of code or the number of discrete tasks/code commits/pull requests completed, all of which can be poor proxies for actual coding efficiency.

Many of the existing coding benchmarks also focus on synthetic, algorithmically scorable tasks created specifically for the benchmark test, making it hard to compare those results to those focused on work with pre-existing, real-world code bases. Along those lines, the developers in METR’s study reported in surveys that the overall complexity of the repos they work with (which average 10 years of age and over 1 million lines of code) limited how helpful the AI could be. The AI wasn’t able to utilize “important tacit knowledge or context” about the codebase, the researchers note, while the “high developer familiarity with [the] repositories” aided their very human coding efficiency in these tasks.

Study finds AI tools made open source software developers 19 percent slower