60 Second AI Update

August 13, 2025

1. GPT-5 Launch: High Hopes, Mixed Reality

GPT‑5 is now live for all ChatGPT users. Promising “PhD-level” reasoning, multimodal capabilities, +256K context windows, and modes like Auto, Fast, and Thinking—yet user backlash is real. Complaints include a colder tone, factual slips, and frustrating usage caps, prompting OpenAI to reinstate GPT‑4o for paid users. (WIRED)

=> This launch highlights a growing tension: AI advancing in capability, but failing in emotional resonance and reliability. OpenAI’s quick rollback and promises of personalization show an evolving understanding of AI as both tool and companion. How they balance innovation with user comfort could define its next era.

2. Manus AI: Hundreds of Agents, One Intelligent System

Manus, a Chinese AI agent, now supports over 100 autonomous sub‑agents working in parallel. Each can independently handle tasks like research, content, or analysis—waving goodbye to single-agent limitations. (The Wall Street Journal, Axios)

=> This could redefine how we delegate tasks to AI—from one assistant to a full suite of mini‑agents. It’s a step toward scalable, distributed intelligence that can handle complex, multi-part workflows with minimal human intervention—a potential game-changer for efficiency.

3. AI Scientist-v2: Fully Automated Research?

AI Scientist‑v2 autonomously generated, submitted, and got accepted a workshop paper at ICLR—no humans involved in the process. (arXiv)

=> This isn’t just automation—it’s self-directed scientific discovery. That’s AI as creator, not just assistant. If this scales, we could see massive acceleration of innovation, for better or worse.

4. DeepCogito v2: Open-Source That Thinks Intuitively

DeepCogito released Cogito v2, an open‑source reasoning AI that internalizes its thinking process using hybrid “intuition” and self‑improvement techniques across four models (70B–671B), all trained for <$3.5M. (Poniak Times)

=> Not only is this cost-effective, but it also brings advanced reasoning transparency to everyone. Open‑source plus efficiency could shift big power to anyone with a GPU—and set a new benchmark for open‑AI innovation.

5. AI Growth Is Slowing—Is AI Hype Ahead of Reality?

A New Yorker analysis questions whether GPT‑5 is a true leap—or just another performance tweak. Critics note diminishing returns in LLM scaling, warning that hype may be outpacing tangible capability. (YouTube)

=> Beware chasing benchmarks at the expense of true advancements. If AI improvements are becoming incremental, market expectations, investments, and regulations will have to recalibrate—possibly ushering in a maturity phase for AI development.

6. "Altman’s Pause": AI’s Moment Before Its Next Surge

The Washington Post argues GPT‑5’s lackluster launch may trigger what they call “Altman’s Pause”—a sociopolitical slow-down where public anxiety and demand for regulation block progress. (washingtonpost.com)

=> This could be a pivotal moment: tech optimism collides with democratic accountability. How institutions respond—through training, infrastructure, and thoughtful policies—might decide whether AI stagnates or thrives.

Why You’re Getting This…

AI is the fastest-moving technology in history—transforming healthcare, jobs, politics, and how we create. 60 Second AI gives you the top stories that matter, in under a minute.