Elon Musk has announced his plans to create another artificial intelligence (AI) system that he calls “TruthGPT.” Musk described the AI system as a “maximum truth-seeking AI” that aims to understand the universe. According to him, such an AI would not pose any threat to humans, as it seeks to understand the world rather than destroy it. This announcement comes as he criticizes OpenAI, the AI research institute where he previously played a significant role in starting it. Musk accused OpenAI of training their models to be politically correct, which he believes to be a way of being untruthful.
Musk has shown concern about AI in the past, stating that it could have too much power to manipulate public opinion or even cause civilizational destruction. He has also signed a letter asking for the temporary pause of training models more powerful than GPT-4 for at least six months.
In recent reports, Musk has been working on recruiting former DeepMind employee Igor Babuschkin to develop his own AI system, which some experts suggest could rival OpenAI and Google's language models. However, it is still unclear whether Musk aims to develop an LLM or use his AI system to research other areas of AI like fairness in training models.
Despite the lack of clarity over Musk's intentions, his status as a billionaire with a track record of making waves in the tech industry could lead to his project, whatever it may be, making a big impact. Interestingly, the trademark “GPT” associated with OpenAI's language models that Musk is allegedly competing with has already been applied for by OpenAI. Therefore, Musk's “TruthGPT” may only be a temporary project name.
Source:
- https://www.foxnews.com/media/elon-musk-develop-truthgpt-warns-civilizational-destruction-ai
- https://techcrunch.com/2023/04/18/elon-musk-wants-to-develop-truthgpt-a-maximum-truth-seeking-ai/