AI models with higher levels of knowledge cannot yet be created. The lack of high-quality datasets and insufficient funding are causing problems.
Developers Anthropic, OpenAI, and Google have set their sights on developing AGI. This advanced form of AI would be as smart as humans. The first company that can make this development a reality will benefit greatly. On the one hand, it has an important power source in its possession that could potentially even be dangerous to humans. On the other hand, it would take an important step in AI development, leading the way for getting new financial resources.
For that to happen, however, the necessary financial resources must first be invested. It has become more difficult to unlock those because the latest developments caused very little commotion.
New data sources needed
At the root of the stagnation in AI development is the lack of unused qualitative data sources. Anthropic CEO Dario Amodei warned of this earlier this week in a podcast discussing AGI development in detail. “We may run out of data,” he warned.
“The AGI bubble is bursting a bit,” Margaret Mitchell, chief ethicist at Hugging Face, told Bloomberg. She frames that statement within the public reaction to the latest developments by AI companies. OpenAI’s release of o1, for example, did not reach the quality level so long hinted at in rumors of “Project Strawberry”.
To get hold of new data sources, developers try to secure exclusive contracts. In turn, those contracts cost money and take time, so developments seem to have been put on hold.
In the podcast, Amodei goes on to venture that AGI will surpass human capability within three years. “If you just look at the speed at which these capabilities are increasing, you think we will be there by 2026 or 2027.”
Also read: Problems with Orion model forces OpenAI to change strategies