Anthropic and OpenAI are making significant advancements in their large language models (LLMs) to enhance coding capabilities and address the needs of enterprises in terms of ethical guidelines, transparency, and data privacy. These technology vendors are continuously updating and developing add-ons for their LLMs to cater to the requirements of various use cases.
OpenAI recently introduced a tool called code interpreter as part of their ChatGPT Plus offering. This tool enables users to analyze data, create charts, edit files, and perform mathematical operations using prompts. OpenAI initially compared the code interpreter plugin to a very eager junior programmer working at the speed of your fingertips when it was introduced in March.
The introduction of such updates and plugins facilitates the crowdsourcing of use cases for these LLMs. OpenAI has encouraged its users to share their most interesting outputs and applications of the code interpreter tool on its Discord server. The use cases shared so far include generating GIFs, converting natural language prompts into C++ code, and analyzing spreadsheets. However, some users have also reported instances where the tool failed to execute certain prompts successfully.
Since the launch of ChatGPT plugins and the code interpreter tool, users have found diverse ways to leverage these tools within online communities. However, it is important to note that these tools are still a work in progress, similar to Anthropic’s Claude. For example, a user illustrated a situation where the code interpreter struggled to combine two images when prompted. On the other hand, the tool has displayed proficiency in converting data sets into graphs, as evidenced by user examples shared on the forum.
Nevertheless, even in cases where these tools typically succeed, users should exercise caution and not solely rely on their outputs. According to Anthropic’s research, Claude models can still produce incorrect information, fabricate details, and fill knowledge gaps with false data. Therefore, they should not be used independently in high-stakes scenarios where erroneous answers may result in harm. Despite this limitation, Claude is already accessible to users through third-party apps and services like Slack.
The focus on improving coding capabilities aligns these generative AI tools more closely with enterprise requirements. This development comes at a time when public interest in these tools appears to be waning. Recent data from internet analytics firm Similarweb shows that ChatGPT’s desktop and mobile web traffic in the U.S. decreased by 10.3% between May and June. Additionally, global website visitors spent less time on the ChatGPT website, with an 8.5% decline.
While the general public’s enthusiasm for ChatGPT and similar tools may be diminishing, technology leaders are inundated with a plethora of AI products entering the market. AI tools currently comprise the top three fastest-growing software products across over 2,000 categories on the software marketplace and review website, G2.
By continuously enhancing coding capabilities and addressing the needs of enterprises, Anthropic and OpenAI are striving to meet the evolving demands of users and stay at the forefront of the AI landscape.
