The AI startup Anthropic has launched a platform for scientists called Claude Science and announced plans to develop treatments for "advanced" diseases. This was reported by The Verge.
The scientific platform was introduced at the event The Briefing: AI for Science. According to the announcement, the solution integrates disparate tools and datasets, as well as creating graphs and visualizations.
According to Anthropic, clients from the biotechnology and pharmaceutical sectors are already using the product.
The Verge noted that OpenAI, Amazon, and Google already offer their own medical tools. However, Anthropic is the first to announce plans for independent drug development, rather than just acting as a software provider.
When pressed by journalists about the specific research directions and types of diseases, the company did not provide answers.
According to the publication, over the past year, Anthropic has been actively hiring biologists and building its own laboratories. Professor Namshik Khan from Cambridge University and co-founder of CardiaTec mentioned that the company reached out to his academic colleagues for recruitment.
He noted that the AI project has also attracted some specialists from AstraZeneca, Novo Nordisk, GSK, and other major research institutions.
AI in Medicine
Experts consulted by The Verge confirmed that artificial intelligence is already being used at various stages of drug development, but it does not replace experimentation and human labor.
Khan listed typical scenarios including the search and optimization of chemical compounds, in-depth research, data analysis, clinical trials, and production.
Professor Matthew Todd from University College London added that "drug development using AI" is too broad a term. In practice, such systems are more likely to accelerate research and help preliminarily test ideas.
Experts believe that there is still a long way to go before any "AI drugs" are approved for human use. They highlighted the challenges of obtaining regulatory approvals and the lack of publicly available high-quality experimental data.
Professor Frank van Delft from Oxford University stated that potential drugs still need to be tested for efficacy, toxicity, and suitability for storage and delivery, despite the assistance of AI. He emphasized that clinical trials are the most expensive part of the process.
It is worth noting that in July, the AIRI research institute introduced the GENATATOR tool based on a neural network for mapping genes according to DNA sequences.