Anthropic, a company known for its forays into advanced artificial intelligence technology, recently announced enhancements to its AI, Claude 3.5 Sonnet, expanding its capabilities to include data analysis. This development positions Claude as a direct competitor to OpenAI's ChatGPT, which already offers a similar Advanced Data Analysis feature.
The new feature allows users to harness the power of Claude 3.5 Sonnet to analyse data by generating and running code. Unlike ChatGPT, which employs Python for its data operations, Claude utilises JavaScript. This is an intriguing choice given Python's established dominance in data science and machine learning. Python benefits from extensive libraries like Pandas and TensorFlow that facilitate data manipulation and analysis, whereas JavaScript is traditionally more web-focused, strong in visualisation rather than sheer data processing capabilities.
One of the significant advantages Claude offers is accessible data analysis for all users, including those on the free version. In contrast, ChatGPT restricts its advanced data analysis features to paid subscribers, putting Claude at an advantage for budget-conscious users seeking to explore data without incurring costs.
Testing the Limits
A practical test conducted by technology journalist David Gewirtz highlights some of the constraints and capabilities of Claude. Gewirtz ran a series of experiments using Claude's free version to gauge its data analysis capabilities. Starting with a comprehensive dataset from the US Social Security Administration that recorded baby names from the years 1880 to 2023, he found Claude unable to handle large files due to data upload size limits.
Each file without a year marker necessitated preparatory processing, which Claude effectively managed by generating a shell script to aggregate data from multiple files into one. However, the consolidated file still exceeded the upload capacity of Claude's platform. Instead, Gewirtz further tailored the data to significantly reduce its size, achieving a manageable file that still maintained comprehensive yearly name records.
Running analyses on this dataset enabled him to observe the AI's performance in generating visual data outputs but revealed limitations in rendering clear data labels on charts. Despite these obstacles, Claude managed the parsing and visual representation of the smaller dataset successfully, with some manual intervention required due to technical restrictions.
Comparison with ChatGPT
When compared to ChatGPT, Claude demonstrates both strengths and weaknesses. ChatGPT managed significantly larger datasets effortlessly, processing tens of thousands of records in various formats without the constraints observed in Claude. This contrast signifies a meaningful gap between the two AI models when handling extensive data collections.
Gewirtz also explored Claude's limitations concerning computational capacity, noting that the free version permitted only up to five file imports and imposed constraints on chat interactions. The Pro version promises to extend these capabilities by five times, yet empirical tests indicate even then it might fall short of ChatGPT's extensive processing ability.
Conclusion from the Tests
Although Claude 3.5 Sonnet's introduction to the realm of data analysis provides an interesting avenue for non-commercial users to explore data processing without incurring costs, its efficacy in handling large sets remains questionable. The current capacity limitations make it appropriate for small-scale data projects rather than comprehensive analytical tasks which require handling large volumes of information.
Anthropic's choice to deploy JavaScript for data analysis rather than Python results in constraints, particularly visible in its graphical data presentation capabilities, failing to exploit JavaScript's potential in data visualization fully. Despite these limitations, the potential for Claude’s future enhancements remains promising, particularly with ongoing technological refinements.
Anthropic has been contacted for comment on these findings. As the landscape of AI data analysis continues to evolve, both users and technology observers are keenly watching developments from both Anthropics and competitors like OpenAI in this burgeoning field.
Source: Noah Wire Services