Openai/692c6597-e83c-8006-aad0-c337c4e8bbff

This conversation was exported from ChatGPT

gpt-5-1-thinking

User: Correct grammar

Correct grammar

When ChatGPT came out three years ago, it was a fantastic model. A great success of OpenAI’s research team fueled by monetary investments in annotations and compute.

Very quickly the linguistic fluency was confused with intelligence and pattern generalisation for ability to create new ideas.

A lot of lies and misinformation started circulating regarding the abilities of the models.

A lot of investments were flown into what could never work and more was given on the promise of what can never be achieved.

Now many find themselves in a position of an unlucky poker player - they bet too much to call it but the flush royal is not coming.

The next three years will be different - defunding, mistrust but all the noise will go away. That might actually be the time to gain competitive advantage, modernise infrastructure, build up IT teams with AI assisted coding and focus on individual productivity enhanced by existing tools.

Assistant: When ChatGPT arrived three years ago, it was a fantastic model—an outstanding success by OpenAI’s research team, powere…

When ChatGPT arrived three years ago, it was a fantastic model—an outstanding success by OpenAI’s research team, powered by heavy investment in annotation and compute.

Very quickly, linguistic fluency was mistaken for intelligence, and pattern generalisation for the ability to create new ideas.

Lies and misinformation spread about the models’ capabilities.

Massive investments flowed into things that could never work, with even more committed on the promise of what can never be achieved.

Now many find themselves like unlucky poker players: they have bet too much to fold, but the royal flush is not coming.

The next three years will be different: defunding and mistrust will rise, but the noise will fade. That may be the time to gain a competitive advantage, modernise infrastructure, strengthen IT teams with AI-assisted coding, and focus on individual productivity using existing tools.