New breakthroughs in the neural capabilities of artificial intelligence (AI) systems have emerged as a potential solution to the problem of AI hallucinations, which has been a major challenge in the industry.
AI hallucinations occur when large language models are trained as generative AI systems and produce new but incorrect information that is not present in their original data set. However, the recent discovery of neural features could offer a way to address this issue, as stated by Dario Amodei, the co-founder and CEO of Anthropic.
Amodei expressed his views on this matter during a speech at Viva Tech Paris 2024, explaining that the challenge with hallucinations lies in the fact that AI protocols are unable to discern the accuracy of their output. Nevertheless, with the aid of the newly discovered features, AI hallucinations could become a thing of the past.
Anthropic has already been conducting research on modifying internal features of its AI assistant to alter its behavior. Through this process, they developed a demonstration model that caused the AI to focus intensely on the Golden Gate Bridge. Anthropic shared this information in a post on May 23.
Earlier in May, Anthropic, a prominent AI development firm, launched its AI chatbot named Claude in Europe. Since May 14, Claude has been accessible through its web application and iOS app in multiple European languages, including French, German, Spanish, and Italian.
Amodei also emphasized that the primary obstacle to AI development is no longer the scarcity of data, thanks to the creation of synthetic data sets. Instead, the main challenge lies in the shortage of skilled developers. While synthetic data shows promise for AI scalability, Amodei cautioned against an excessive reliance on it, as an infinite ratio of synthetic to real-world data would result in weaker and less realistic data sets.
Investor interest in the AI industry has been steadily increasing since the beginning of 2024. On May 27, Elon Musk’s company secured a $6 billion Series B funding round, with the aim of positioning its Grok chatbot as a competitor to ChatGPT by the end of 2024.
In other news, a Hong Kong streaming firm has announced its plan to purchase $100 million worth of cryptocurrency, while Worldcoin has faced sanctions, as reported by Asia Express magazine.