ChatGPT Now Handles Complete Workflows Autonomously
OpenAI launches autonomous ChatGPT agents for multi-step tasks, new research reveals AI coding tools make experienced developers 19% slower, Google adds advanced search capabilities, and neuroscience-backed methods for restoring focus.
Artificially Intelligent Tuesday, July 22, 2025 (Audio Narration with Commentary)
0:00
/709.360136
🤖 ChatGPT Now Handles Multi-Step Tasks Using Its Own Computer
What it is: ChatGPT agent is OpenAI's new autonomous task execution system that combines web browsing, analysis, and action capabilities in a single interface. Unlike previous AI assistants that only respond to questions, this system can complete multi-step workflows independently using its own virtual computer.
Key capabilities: The agent can navigate websites, analyze data, create presentations and spreadsheets, manage calendar appointments, conduct research, and execute complex tasks from start to finish. Users maintain control through permissions and can interrupt or redirect tasks at any point. Pro users get 400 monthly messages, while Plus and Team users receive 40 messages per month.
Why it matters: This represents a shift from AI as a conversational tool to AI as a task executor. Instead of breaking complex work into multiple prompts, you can delegate entire workflows—like "research three competitors and create a presentation" or "analyze this data and generate a report." The agent handles the coordination between different tools and maintains context throughout the process, eliminating the need to manually guide each step.
🧠 AI Coding Tools Slow Experienced Developers by 19%
What it is: METR conducted a randomized controlled trial with 16 experienced open-source developers working on real projects. Each developer completed 246 tasks (averaging two hours each) on repositories they knew well, with tasks randomly assigned to either allow or prohibit AI tools like Cursor Pro with Claude 3.5/3.7 Sonnet.
Key findings: Despite developers predicting AI would speed them up by 24%, they actually took 19% longer to complete tasks when using AI tools. This slowdown persisted even though developers could choose when to use AI and reported believing AI had helped them by 20% after the study. The effect was consistent across different analysis methods and wasn't explained by experimental artifacts.
Why it matters: This challenges the widespread assumption that AI coding tools automatically boost productivity for skilled developers. The study reveals a significant gap between perception and reality—both developers and external experts dramatically overestimated AI's impact. For practitioners, this suggests being more critical about measuring actual time savings versus perceived benefits when adopting AI tools in your workflow.
🔍 Google Search Gains Advanced Research and Business Calling Capabilities
What it is: Google Search's AI Mode, available to Google AI Pro and AI Ultra subscribers, now includes access to more powerful AI models and automated research tools within the search interface.
What's new: Two significant capabilities have rolled out to subscribers in the US. First, Gemini 2.5 Pro is now available in AI Mode, offering improved performance on complex reasoning, mathematical calculations, and coding questions compared to the default model. Second, Deep Search can now conduct comprehensive research by automatically running hundreds of searches, analyzing disparate information sources, and generating fully-cited reports in minutes rather than hours of manual research.
Additionally, Google has introduced AI-powered calling for local business inquiries. When searching for services like "pet groomers near me," users can select "Have AI check pricing" and Google will call businesses directly to gather pricing and availability information, then present consolidated results without requiring phone calls.
Why it matters: These features address two common productivity bottlenecks in knowledge work. Deep Search eliminates the tedious process of cross-referencing multiple sources for complex research tasks, while the business calling feature removes the friction of gathering basic service information. For professionals who regularly conduct market research, competitive analysis, or need to coordinate local services, these tools can recover significant time previously spent on manual information gathering.
More AI everywhere doesn’t mean better results. Small teams of AI experts beat mass licenses. Attackers now use AI to run cybercrime at machine speed. Leaders can’t keep up. Tools still forget us. The winners? Focused expertise and hybrid AI, not AI everywhere.
You've built something and you need to know if it works. So you do what's sensible—you ask an LLM to grade it. Factual accuracy, code quality, agent outputs. The machine judges the machine, and you get a number you can act on. Except that number
Imagine you have a bunch of teams, some with AI, some without, and some where everyone gets their own AI. Researchers ran a big experiment with over 400 people to see what actually happens when you mix and match humans and AI in different ways.
Here’s what they found:
