Artificial intelligence (AI) could be in a cul-de-sac thanks to a lack of data and technological restrictions, according to a report from Bloomberg.
It’s been almost three years since generative AI came on the scene in the form of chatbots, AI image generators and music generators. Such a big leap forward has caused many people to wonder what will happen next.
So did excited tech companies and their shareholders, but Bloomberg reports that Google, OpenAI and Anthropic are all struggling to build more advanced AI.
According to Bloomberg sources, the OpenAI model is working, Orion, is not hitting internal expectations. Likewise, Google’s newest iteration of Gemini isn’t much better than the previous one either. Anthropic has also delayed the release of its Claude model.
One reason cited is that “it has become increasingly difficult to find new, untapped sources of high-quality, human-made training data that can be used to build more advanced AI systems.”
This is really fascinating. It is well known that AI companies are ripping almost all the data available on the open web to build different models. It’s safe to bet that almost every photo online has been taken for AI training purposes.
It’s something I wrote in June 2023 in which I pointed out that AI image generators cannot survive without fresh photography.
Bloomberg says that with all the open web scraped, tech companies are having a hard time closing the gap. Some of them have turned to AI imaging, but researchers have found that this method has limitations. A study revealed that AI trained on computer generated material turns to mush.
All this puts a dampener on the dream of artificial general intelligence (AGI). This refers to supposed AI systems that are smarter than humans. OpenAI and Anthropic have previously stated that AGI is close.
“The AGI bubble has kind of burst,” said Margaret Mitchell, chief ethics scientist at AI startup Hugging Face. Bloomberg. Mitchells says AI companies need to take “different training approaches” to have AGI.
“It’s less about quantity and more about quality and diversity of data,” adds Lila Tretikov, head of AI strategy at New Enterprise Associates and former deputy chief technology officer at Microsoft. “We can generate quantities synthetically, but we struggle to get unique, high-quality datasets without human guidance.”