Microsoft Research 播客第 98 期汇总编程智能体降本研究与 KLS 猜想解法等进展
Microsoft Research Focus – Edition #98
- 1Microsoft Research 第 98 期播客汇总语音翻译、KLS 猜想、编程智能体提效与多语言分词器研究进展
- 2研究分析 1350 万次编程会话,不到 8% 需要上下文压缩,却占总 AI 处理成本的 44%
- 3配套 APEC 方法在后台预生成摘要,评估中编程表现提升,成本最多降 25.4%、执行时间降 9.6%
- 4悬置 31 年的 KLS 猜想被解决,官方称结果经独立验证并用计算机辅助方法完成形式化验证
这期研究汇总的核心信号是,AI 编程智能体的成本大头集中在触发上下文压缩的少数长会话上,该环节已出现可量化的优化方向。
- 影响:依赖 AI 编程工具的企业,少数长会话可能贡献近半处理成本,预算可按会话长度拆分核算;摘要预生成思路能把压缩开销移出工作流,但相关方法仍属研究阶段成果,短期需经现有产品功能间接受益。
- 参考:可考虑统计自家 AI 编程工具的会话长度与成本分布,定位触发上下文压缩的长会话;评估智能体采购时,可把上下文压缩的计费方式与效率表现列入询价清单。
- 依据:1350 万次会话统计与 APEC 最多 25.4% 降本、9.6% 提速均为微软研究方口径,暂无独立评测;KLS 猜想解法官方称经独立验证并以计算机辅助方法完成形式化验证。
What’s new in research?
In this edition of Microsoft Research Focus, based on the Research Focus newsletter (opens in new tab), host Anson Ho shares some of the latest news and cutting-edge research from across Microsoft—from bringing AI-powered universal translation closer to reality and solving a 31-year-old mathematical challenge to making AI coding agents faster and more efficient.
Catch up on the ideas, tools, and discoveries researchers are exploring today and get a quick look at what’s coming next.
Related links:
- A research intern’s journey from computer vision to physical intelligence at Microsoft Research Asia – Tokyo
- Breaking the Language Barrier — How AI Speech Translation Is Rewriting Our Digital Lives
- Solving a 31-year-old grand challenge in mathematics – the KLS conjecture
- Benchmarking Long-Horizon Agent Context Compaction with CompactionBench
- Latent Core Tokenizer: Compress, but Meaningfully
Chapters
0:35: Internship: Growing expertise, building community (opens in new tab)
1:28: Breaking the language barrier with AI (opens in new tab)
2:10: Solving a 31-year-old mathematics challenge (opens in new tab)
2:56: Making AI coding agents more efficient (opens in new tab)
3:57: Improving AI tokenization across 104 languages (opens in new tab)
Subscribe to the Microsoft Research Podcast:
Transcript
[MUSIC]
This is Microsoft Research Focus, your quick look at news and cutting-edge research from across Microsoft.
I’m Anson Ho, and these are some of the latest advances we’re tracking.
Coming up in this edition: how AI is bringing the possibility of a universal translator closer to reality; a breakthrough that solves a 31-year-old mathematical challenge; a new approach to make AI coding agents faster and more efficient; and more.
Here’s what’s new.
[MUSIC FADES] At the Microsoft Research lab in Tokyo, research intern Lilika Makabe found opportunities to grow her expertise while helping to shape a new research community. Makabe joined the lab with a background in computer vision and worked alongside researchers exploring robotics and physical intelligence, AI systems that can interpret information and act in the physical world. Her experience highlights how research internships can offer more than technical training. Through hands-on experimentation, interdisciplinary collaboration, and everyday conversations with colleagues from different backgrounds, Makabe expanded her research perspective while contributing to the lab’s growing community. You can read about her experience on the Microsoft Research blog. [TRANSITION MUSIC] For decades, science fiction has imagined a world where language barriers no longer stand in the way of communication. For example, in 1967, Star Trek introduced the fictional Universal Translator. Now, advances in AI are bringing that possibility closer to reality. In a new conversation, Microsoft Applied Science Manager Jinyu Li explores how AI is advancing the prospect of communication across languages, the progress researchers have made, and the challenges that remain. To learn more, catch the full episode of the podcast Our Digital Life, available to listen or watch on YouTube and podcast platforms. [TRANSITION MUSIC] For the last 31 years, the KLS conjecture has been one of the central open problems in high-dimensional geometry. But not anymore. New research has resolved the conjecture. This removes a major theoretical obstacle to understanding the efficiency of randomized algorithms used across science and computing. The breakthrough establishes a universal principle for log-concave distributions: simple flat cuts can identify their narrowest bottlenecks within a fixed factor, regardless of dimension. The results have been independently validated, and the proof has also been formally verified using computer-assisted methods. Learn more about this breakthrough on the Microsoft Research blog. [TRANSITION MUSIC] Coding agents can help when writing and modifying code, but long sessions can max out the amount of information an AI model can process at once. As a result, coding agents will summarize earlier interactions in a resource-intensive process known as context compaction. A new study looked at 13.5 million coding sessions and found fewer than 8% required context compaction. Yet those sessions accounted for 44% of total AI processing costs. To address this challenge, researchers developed COMPACTIONBENCH. COMPACTIONBENCH is a benchmark for evaluating context compaction during long-running coding tasks. They also introduced APEC, a method that prepares summaries in the background instead of interrupting an agent’s work. In evaluations, APEC improved coding performance while reducing costs by up to 25.4% and execution time by 9.6%. [TRANSITION MUSIC] Before AI models can process text, they break it into smaller units called tokens. How that text is divided can affect both efficiency and performance, especially across different languages. Most tokenizers are designed to compress text as efficiently as possible. But a new study suggests that compression alone isn’t enough. Researchers introduced the Latent Core Tokenizer, or LCT, which identifies reusable components of words before building a shared vocabulary across languages. In evaluations spanning 104 languages, LCT produced more linguistically meaningful tokens than leading alternatives while maintaining comparable efficiency across languages. It also improved model performance across four multilingual benchmarks. The findings suggest that better tokenization depends not just on how much text can be compressed, but on how well the underlying structure of language is preserved. [MUSIC] And that’s your Microsoft Research Focus. From the research culture and the collaborations driving discovery to producing more linguistically meaningful tokens, these are just a few of the developments researchers are exploring across Microsoft. Want to learn more and stay connected? Explore the latest research, discover the people behind it, and keep tabs on the ideas shaping what’s next by visiting aka.ms/MicrosoftResearch (opens in new tab) and follow Microsoft Research on all major social platforms. I’m Anson Ho, and thanks for tuning in. We’ll see you at the next Microsoft Research Focus. [MUSIC FADES] Internship: Growing expertise, building community
Breaking the language barrier with AI
Solving a 31-year-old mathematics challenge
Making AI coding agents more efficient
Improving AI tokenization across 104 languages
The post Microsoft Research Focus – Edition #98 appeared first on Microsoft Research.
来源:Microsoft Research · microsoft.com