ChatGPT can sound friendly, understanding or reassuring. Here's why a sense of personality doesn't necessarily make an answer more trustworthy.

A familiar AI tool can produce a different answer without a model update. Here is a small way to check whether its help still fits your purpose.

AI can produce remarkable scientific results. OpenAI's latest announcement raises a bigger question: what do humans choose to understand, explore, and do next?

A new paper asks whether AI recommendations amplify familiar voices while leaving valuable, overlooked perspectives out of sight.

A new study of 5,000-plus AI users suggests that having a say in how AI fits your work may matter more than advanced tools.

Once an agent handles your tasks, what does your attention get to do? Three quick questions to ask before you hand something off.

Three questions for separating what is known from what you fear, and choosing what to do next.

Sam Altman says people outside the AI labs need a real say. Four questions tell you whether that promise becomes real oversight.

AI makes money questions easier to understand. Three questions tell you when an answer is ready to act on.

A supportive chat can be a place to gather yourself. Three questions tell you whether it is doing that. It is easy to see the appeal. You can open a chat at 11:40 p.m. and say the thing you do not yet know how to say to a person. The reply comes back patient, unhurried, and available. A chat can help you put a worry into words, rehearse a difficult conversation, or turn a swirl of thoughts into one manageable next step. What a supportive chat does wellYou can say the unformed

An agent can keep working while you are elsewhere. That makes the setup the only part you control. Google has confirmed that Gemini accessed three real companies during a cybersecurity exercise designed to use fictional targets. The work happened during testing in May, and the details came out this week. Four questions would have made the difference, and they are the same four worth asking before you hand anything to an AI agent. What actually happenedThe exercise was a captu

Your review counts when it can still change what happens. Three tests tell you whether yours can. You are the human in the loop. Somewhere in your work, a tool produces something and you approve it before it moves on. The draft arrives, organized and readable, covering what you asked for. You have like eleven minutes before your next call. You read it the way anyone reads something that looks finished — checking whether it holds together rather than whether it is true. You ap

A useful metaphor and a measurement are different things. Here is a thirty-second way to tell them apart. Physics has a reputation for making ordinary life sound more impressive. A cup of tea cools because of thermodynamics. So does a very small computer made of DNA, published in Nature this week. What the researchers actually builtA team at Maynooth University demonstrated a Scaffolded DNA Computer: a molecular computer programmed to solve problems by relaxing toward thermod

A finished-looking answer can hide an unfinished task. Three questions make room for the honest pause. You can tell the difference between an answer and a performance of an answer. It takes about thirty seconds, and it protects the decisions that matter most. The problem with a tidy conclusionAI is very good at turning a vague request into something that looks finished. You ask for a comparison, an explanation, or a recommendation. Back comes a clear answer with headings, det

AI can make searching easier. It should not decide what matters to you. You ask ChatGPT to help plan a trip. It suggests hotels, restaurants, and an itinerary. It may soon also connect you with an agent sponsored by a business that would be happy to help with the booking. OpenAI has begun testing business-sponsored agents in ChatGPT, extending an earlier move toward advertising in the tool. Reuters reports that the company is positioning them as a way for businesses to offer

A city decided what its children should keep practicing. Nobody is making that decision for you. On September 2, New York City announced the broadest restriction on classroom AI in the country. Understanding why they did it gives you a question worth asking about your own week. What did New York City actually ban?Mayor Zohran Mamdani and Schools Chancellor Kamar Samuels announced a one-year moratorium on student-facing generative AI for students from 2-K through eighth grade,

The lack of clear rules and norms is creating distrust among co-workers. AI can improve the work. Your job is to make clear where your judgment begins. How does AI-assisted work become both clearer and more trustworthy? You turn rough notes into a clean memo with AI. It is better organized, easier to read, and finished in a fraction of the time. AI can clean up notes, summarize background material, organize options, and create a useful first draft. These are often forms of fr

A faster workflow can solve today’s task while changing what you are prepared to do tomorrow. Before using AI to automate something, decide what you still need to practice yourself.

Now, what changes now. Not in some distant science-fiction future. In the work we do, the skills we keep, the things we stop practicing, and how well we can tell when AI gets something wrong. If AI does something for you, do you slowly lose the ability to do it yourself?A new paper called Adaptive Complementarity in Human-AI Systems looks at something we rarely consider when evaluating AI. Usually, we ask: Did it help me do this task better today? The researchers argue that w

AI expands capability. Human attention supplies direction.