It’s gonna happen. Sometime in the next year, an executive from OpenAI, Anthropic, Google, or Meta will be standing on a brightly lit stage, in front of a screen full of benchmark scores, and claim that their latest AI model is proof that they have achieved Artificial General Intelligence (AGI). I fear that the news media will not be ready for this moment.

First of all, there is no technical, universally agreed upon, provable threshold for AGI. The most common description is an AI system that can outperform humans in every area. There are lots of benchmarks out there that claim to measure how well these models perform in different areas of knowledge, but they have their flaws.

Here are some things journalists should keep in mind when reporting on this imminent declaration of AI superiority.

Don’t anthropomorphize AI. If the latest model from one of these companies gets a high score on the bar exam, all that means is that the model knows how to answer questions on the bar exam (and it likely has already seen the answers). That doesn’t make the AI model smarter than a lawyer.

Be wary of the language used by AI companies when describing their products. For example, “reasoning” models don’t actually reason or ponder, they just break queries down and spend a lot more time (and energy, and water) to process the query. It does yield better results, but it’s not thinking. Remember, these are probabilistic systems that are very good at pattern matching for human language.

Everyone is making their own yardsticks. Companies like Google, Meta, OpenAI, and Anthropic will boast about their high scores in benchmarks like ARC-AGI, Humanity’s Last Exam, Vending-Bench 2, and where they rank on leaderboards like LMArena — which is determined by human-based blind-choice side-by-side taste tests. The AI companies are also making their own benchmarks, and sometimes might be putting their thumb on the scale when testing. These tests should be scrutinized and explained to readers. There really isn’t an independent body with truly universal standards that can fairly measure the performance of all these models.

Beware of the hype…and the anti-hype. There’s little doubt that the tech industry finds itself in a bubble of some size and shape, and it is inflated by endless hype from very powerful people who have a lot of their money at stake. The Trump administration is clearing the way for AI companies to get everything they want with almost no oversight — like quick approvals of nuclear power plants for data centers, access to government data for training, juicy federal contracts, and maybe even a future bailout if and when the bubble pops.

If you can resist the hype, also try to take a critical view at claims that generative AI is a total con. The generative AI boom has indeed yielded real productivity gains and actual revenue for some applications, like software development and customer service. It is a legitimately important technological breakthrough. I have seen this myself using AI-assisted coding, and it really has been a positive boost for my creativity. But it has not completely changed the world for everyone, nor is it likely to just because Sam Altman, Elon Musk, or Mark Zuckerberg says it will.

Focus on how AI is affecting people. No matter how many times tech executives show a demo of how they think people might use their products, people are going to use it in good and bad ways they haven’t even considered. We’ve already seen too many disturbing stories of people suffering mental health crises turning to AI chatbots for help, only to have it harm them. Lots of good journalism has been done in this area, and there should be more.

As more of the public embraces AI in their daily lives, we should ask people what they think AI is actually capable of doing, and explain what it can actually deliver.