Is AGI Here? A Reality Check on GPT-4o and Friends
Everyone's talking about Artificial General Intelligence (AGI), but are we there yet? We cut through the hype to show what today's AI can actually do, and where it consistently fails.
May 2, 2026 · 1 min read · SuperThinking team
No. AGI is not here.
It’s not even close. The slick demos showing a cheerful AI voice instantly solving problems by looking through a phone camera are impressive, but they aren't proof of general intelligence. They are proof of incredible engineering and unbelievably large datasets.
We get excited because these models are finally good at things that feel uniquely human, like conversation and visual interpretation. But confusing a good simulation with the real thing is a trap. Let's ground the conversation in what these models can—and absolutely cannot—do today.
What Models Can Do (That Feels Like AGI)
Let’s be clear: today's frontier models are miraculous. GPT-4o, Claude 3 Opus, and Gemini 1.5 Pro can perform tasks that were science fiction three years ago. If you haven't tried it, paste a screenshot of a website UI into one of these models and ask it to write the React code. It works. Frighteningly well.
They can:
- Synthesize Information: You can upload a 200-page PDF of earnings reports and ask,