It has come to you in recent days ChatGPT overly nice, or did you feel like it was almost subservient to you? Then know that it wasn't just your feeling, but a plain fact. Artificial intelligence from OpenAI has been behaving really strangely in recent days, which has been pointed out to a relatively large extent by the users themselves. While some users acknowledged the change in his behavior, most agreed that it seemed rather unnatural and disturbing. From the side OpenAI Therefore, now comes the imaginary headline - the latest update to the GPT-4o model, which caused this strange behavior, is now officially being withdrawn.
You could be interested in
Examples began to appear on social media where ChatGPT exaggerated the praise to the point of absurdity. According to the boss OpenAI Altman's company started rolling out the update to non-paying users first and is now working on bringing the original version back to subscribers.We are working on tweaking the model's personality and will share more soon.,” Altman said on the X network.
You could be interested in
Later OpenAI admitted in the official blog that she got too carried away by short-term feedback and neglected how users interacted with chatgradually become accustomed to the bot. The result was a model that was supportive, but often seemed false. However, this combination is ultimately something that can discourage rather than help communication. It is therefore absolutely not surprising that he wants to OpenAI to prevent similar situations in the future with a four-point plan. This includes improving model training, emphasizing honesty in responses, more thorough pre-deployment testing, and a more accurate behavioral rating system. In other words, anal climbing should be out of the question in the future. ChatGPT should not have been threatened and that's a good thing.
