高性能なAIモデルを使って自分のAIモデルを学習する手法「蒸留」は、する側はコストを抑えて最先端モデルに近い性能を実現することができますが、される側は不要な負担を強いられるなどのデメリットがあります。こうした蒸留はしばしば「攻撃」と見なされ、対策が行われています。新たに、Anthropicが望まない蒸留を大幅に低減する手法を公開しました。

続きを読む...

AIが数学の未解決問題を次々と解決できるようになる中、著名な数学者のテレンス・タオ氏がAI時代における数学研究の新たな方向性として、「Math 2.0」と題した公開講演をカリフォルニア工科大学で2026年10月9日に行い、その資料を翌10日に自身のブログで公開しました。タオ氏は従来の数学研究を「Math 1.0」と位置付け、AIによって数学の証明が大量に生み出される時代に向けて、研究の目的や評価基準を見直す必要性を説明しています。

続きを読む...

「フォートナイト」などを開発するゲーム会社のEpic Gamesが、ディズニーと共同で開発しているゲームの発売を延期し、AIをもっと活用して開発を強化するよう指示を出したと伝えられました。

続きを読む...

アメリカのAI企業であるSubquadraticが、Jevの12倍に相当する38万4000トークンの入力に対応した意思決定モデル「SubQ Snap」を発表しました。

続きを読む...

Microsoftが意志決定モデル「Microsoft-Decision-1」を現地時間の2026年10月9日にリリースしました。ゲームのレビューを分析するタスクではGPT-6 Solと同等の品質で14倍以上高速、コストは200分の1以下だったと報告されています。

続きを読む...

SpaceXSI(旧称：SpaceXAI)がSNS「X」、AIサービス「Grok」、コーディングAIサービス「Cursor」の有料サブスクリプションプランを統合したことを発表しました。新たなサブスクリプションの名称は「The X Subscription」で、個人向けプランは「Free」「Plus」「Premium」「Super」「Ultra」の5段階に分かれています。

続きを読む...

MicrosoftがカスタムAIチップの第3世代にあたる「Maia 300」を2027年に導入予定であることを発表しました。Microsoftによれば、MaiaファミリーはMicrosoft製品およびサービスの主要なAI推論システムになる予定だとのことで、NVIDIAへの依存が軽減されるものとみられます。

続きを読む...

MetaがAIモデル「Muse Wonder」と「Muse Wonder Full」を現地時間の2026年10月12日に発表しました。どちらのモデルも各種ベンチマークテストでOpenAIやAnthropicの最先端モデルを上回るスコアを記録しています。

続きを読む...

An anonymous reader quotes a report from Bloomberg: Epic Games delayed the release of an upcoming video-game collaboration with Walt Disney and told its developers to use more artificial intelligence tools to finish the game, according to people familiar with the project. The title had been slated for release in March, but is now delayed with no new date communicated, said the people familiar, who asked to not be identified because they weren't authorized to speak publicly. Developers on the game, a shooter featuring characters from the Marvel and Star Wars franchises, were instructed to implement an "AI first" production model using tools such as Anthropic's Claude. Humans will then manage and guide that work. "No more hand-written code," one of the game's directors told staff in an internal message reviewed by Bloomberg. The director added that all art for the game would be "generated with full AI workflows." Even before this announcement, the maker of the popular Fortnite game had been pushing hard for the use of AI across all of its divisions. But this is the first time a major video-game company has attempted to build a title primarily with AI. "We're embracing AI-accelerated development, led by Epic developers who are the visionaries and creative drivers for everything we do," said Jake Jones, communications director at Epic Games, in a statement to Bloomberg News.

Read more of this story at Slashdot.

「Nerf Bench」はAIモデルがローンチデー(サービス開始日)からどれだけ性能を維持しているかを評価するベンチマークです。

続きを読む...

Meta CEO Mark Zuckerberg says the company has reached the AI "frontier" with its new Muse Wonder model and will be offered as a closed, paid product for businesses. Zuckerberg also rejected calls for an industry-wide slowdown, arguing that AI can be managed with conventional guardrails as Meta continues spending heavily on models, infrastructure, and talent. The Financial Times reports: Speaking with the FT as Meta releases its Muse Wonder AI model on Monday, the billionaire chief executive claimed the new system was "going to allow much more sophisticated behaviours that are at the frontier of what is being done in the AI field." [...] "Getting to this point both required a large amount of compute and investment, and a lot of very rigorous scientific work by the group working on it over an extended period of time," said Zuckerberg. [...] Meta said Muse Wonder outperforms Anthropic's Claude Opus 5.5 and Fable 5.1 and OpenAI's GPT-6 Astra on some technical benchmarks. It said the new model can be used to carry out complex tasks from "understanding a lease to making sense of financial or health information." However, Muse Wonder will also be "closed," with its best versions sold directly to businesses. That pits it against Anthropic and OpenAI for corporate AI contracts in a serious test of its ambitions. Zuckerberg signaled further spending as Meta was "already working on the next step up," adding that its model development would "probably feel like it'll go quite a bit faster from here." However, he also shrugged off mounting fears that the technology was advancing at a rate that will soon surpass human control. [...] The Meta chief declined to be drawn on his figure for "P(doom)" -- a much-discussed Silicon Valley term to describe the probability that AI will lead to mass destruction. Instead, he argued that responsibility should fall on the individual lab to ensure model safety. "I'm much closer to the belief that AI is a technology that can be engineered with normal guardrails and built responsibly than the anthropomorphic vision of AI, which is that it is just going to go do its own thing," he said. Zuckerberg has previously rejected calls by the heads of AI labs such as Anthropic's Dario Amodei and OpenAI's Sam Altman for an industry-wide "slowdown" in AI development. He also played a key role in crafting US President Donald Trump's White House "Accord on Super Intelligence," setting out principles for self-regulation of the industry, which critics say falls far short of a formal global AI governance regime.

Read more of this story at Slashdot.

A study of more than 700 software firms found that AI coding agents significantly increase the amount of code developers produce, but not the amount of software companies actually ship. Ars Technica reports: A recent study (PDF) of actual coding practices across hundreds of firms finds that human code review forms a significant "bottleneck" for the overall efficiency of AI coding tools, resulting in "little evidence that firms increase software output or reduce employment" by using them. Any efficiency increased during the actual coding phase, the study authors find, is "absorbed by downstream constraints in the production process"; as "the code review process significantly increases in length, pull requests are more likely to require revisions, and reviewers leave more comments." [...] In terms of raw code being produced, the results are clear and stark. The introduction of AI coding agents at a firm leads to a 30 percent increase in total lines of code generated, a 20 percent rise in the number of total commits, and a 23 percent increase in pull requests on average, the researchers write. But all that extra code doesn't translate directly into improved software output on the firm level. On the contrary, the resolution rate for Issues and Epics (i.e. wholesale software features) tracked by tools like Jira did not change in a statistically significant way after AI tools were introduced (the researchers also found no "compositional shift" in the size or complexity of those Jira-tracked issues across the AI introduction). The reason for that discrepancy can be found directly in the code review process, which takes markedly longer on average after the introduction of AI coding agents. Overall, the average "review process" time between a pull request getting submitted and it being merged into the codebase balloons 49 percent on average after AI agents are introduced. That effect can be seen in more granular data, too, with "the share of pull requests with changes requested nearly doubl[ing], and the number of comments per pull request increas[ing] by 35%" following the AI agent shift, the researchers write. In response to this change, the researchers found a 14 percent increase in the share of workers performing code reviews after AI agents' introduction. They also write that they "cannot attribute significant employment changes to AI" after looking at total active workers across Jellyfish and cross-referencing with LinkedIn data at those firms. While AI could also theoretically help with this review process, the researchers found that, so far, that impact has been marginal. Although 80 percent of measured firms used some form of AI code review by March 2026, AI agents were only responsible for 23.3 percent of all review comments and 10.8 percent of all pull requests, suggesting humans were still responsible for the vast majority of this work.

Read more of this story at Slashdot.