Sign up to save tools and stay up to date with the latest in AI
bg
bg
1

AI Appears to Rapidly Be Approaching Brick Wall Where It Can't Get Smarter

Jun 09, 2024 - futurism.com
Researchers warn that AI companies like OpenAI and Google are quickly running out of human-written training data for their AI models, which could hinder the development of smarter AI. The amount of text data AI models are trained on is growing at a rate of 2.5 times a year, while computing is growing four times a year. This could lead to large language models running out of fresh data by 2026. The trend of using publicly available online data has led to copyright infringement lawsuits, and as companies invest more in AI and lay off workers, the stream of new content could dwindle.

In response to the potential data shortage, AI companies are considering training their models on AI-generated data. However, studies have shown that using AI-generated content can degrade the output quality of the models. There is ongoing debate about the severity of this issue, with some arguing that AI algorithms could become more efficient and produce better outputs with less data or computing power.

Key takeaways:

  • AI researchers warn that companies like OpenAI and Google are rapidly running out of human-written training data for their AI models, which could limit the models' ability to improve.
  • This lack of data is seen as an existential threat for AI tools that rely on large amounts of data, often pulled from publicly available online archives.
  • Researchers predict that large language models could run out of fresh data as soon as 2026, and AI companies may resort to training their models on AI-generated data, which has been found to erode output quality.
  • There is ongoing debate about the potential impact of this issue, with some suggesting that AI algorithms could become more efficient and produce better outputs with less data or computing power.
View Full Article

Comments (0)

Be the first to comment!