June 5, 2024
AI hallucinations

New Neural Features Tackle AI Hallucinations: Anthropic CEO

Breakthrough in AI Technology: Addressing AI Hallucinations

Newly discovered neural features in artificial intelligence (AI) systems could be the key to solving AI hallucinations, one of the industry’s biggest challenges. AI hallucinations, an unintended consequence of training large language models, occur when these models generate incorrect information that isn’t present in their data set.

Understanding AI Hallucinations

AI hallucinations are a significant issue because AI models often produce incorrect outputs without recognizing their inaccuracy. According to Dario Amodei, the co-founder and CEO of Anthropic, the newly discovered neural features hold promise for resolving these hallucinations. Speaking at Viva Tech Paris 2024, Amodei explained:

“Modern AI systems have hundreds of millions of neurons… We found 30 million features in our model. Features for security vulnerabilities or bias against certain kinds of groups… We’re able to now intervene in them and change them.”

These neural features could transform how AI systems handle incorrect outputs, potentially making AI hallucinations a thing of the past.

Anthropic’s Innovations in AI

Anthropic has been exploring how altering internal features of its AI assistant can change its behavior. One example highlighted a demo model that made the AI focus intensely on the Golden Gate Bridge. This finding, shared in a May 23 X post, underscores the potential of these neural features:

“The fact that we can find and alter these features within Claude makes us more confident that we’re beginning to understand how large language models really work.”

Launched in Europe earlier in May, Anthropic’s AI chatbot, Claude, is available in several European languages, including French, German, Spanish, and Italian. The company’s ongoing research aims to further refine AI capabilities and reduce AI hallucinations.

The Future of AI Development

Anthropic CEO Dario Amodei emphasized that the future of AI development will hinge on talent rather than data. While the lack of data has historically been a bottleneck, the advent of synthetic data sets has shifted the challenge to the availability of skilled developers. Amodei noted:

“Data will not be a bottleneck as some firms are starting to create synthetic data… AI advancement will mainly come down to talent…”

He also mentioned that although synthetic data is beneficial for scalability, relying too heavily on it could lead to weaker and less realistic data sets.


The discovery of neural features by Anthropic marks a significant step forward in addressing AI hallucinations. As the industry continues to evolve, the focus will be on leveraging these insights to enhance AI accuracy and reliability. With the potential to revolutionize how AI systems function, these advancements could mitigate one of the most pressing issues in AI today.

Image by freepik

Disclosure Statement: Miami Crypto does not take any external funding, or support to bring crypto news to the readers. We do not have any conflicts of interest while writing news stories on Miami Crypto.

Related posts

Startup Showcases Future Decentralized GPU Infrastructure Network to OpenAI, Uber

Harper Hall

Microsoft Unveils Maia AI Accelerator: A Game-Changer in AI

Robert Paul

Elon Musk’s AI Venture Aiming for $4 Billion to Elevate Grok Chatbot

Cheryl  Lee

This website uses cookies to improve your experience. We'll assume you're ok with this, but you can opt-out if you wish. Accept Read More

Please enter CoinGecko Free Api Key to get this plugin works.