Read the First AI Reality Check

SMACNA’s AI Reality Check is a quick monthly update to help you understand what’s happening with AI. Each post will share simple, valuable insights—what’s helpful, what’s overhyped, and how to think about AI practically.

Earlier this month, OpenAI launched its “Tasks” feature, which allows users to create automated tasks that can run later. At first, this might sound like “reminders” that we’ve seen for years, but there are a few key differences: 

1. You can ask in plain language for some complicated tasks, like “Let me know when Gold passes $1,500 an ounce,” which will include web search and other things, or “Every morning, send me an image of dolphins,” where it will create new images every morning. 
2. You can ask it to look at the full range of ChatGPT filetypes, from Excel to sequel databases and more, so it’s very powerful
3. Per above, it can do these tasks based on conditions, not just regular intervals

OpenAI has created a simple agent framework for consumers to get used to the idea of AI Agents without quite so much complexity. Here’s a quick video.

The Good: It is an easy way to offload some mental effort that people are often not great at, like remembering to do things or paying attention to whether something has happened.

The Bad: They’re still going to be narrow software, so they will miss things and send irrelevant things sometimes. Offloading mental effort is great, but we’re going to need to learn when not to fully rely on still-limited systems.