Year in review 2025: AI in data science [Python/R] (posit.co)

In 2025, the AI and data science landscape saw considerable advancements, with significant model releases shaping the sector. Notably, OpenAI launched GPT-5.2 to mixed reviews, highlighting a trend of unpredictability in newer models. Meanwhile, Anthropic introduced Claude Code and Claude 3.7 Sonnet, establishing the "coding agent" concept as a dominant theme, later echoed by OpenAI and Google. DeepSeek's open-source reasoning model R1, released earlier in the year, stirred market anxiety regarding global competition in AI, causing a selloff in related stocks. Amid these developments, Posit rolled out innovative tools like chatlas and Databot, which catered to Python users and facilitated exploratory data analysis, signaling a keen focus on enhancing coding capabilities. This year's achievements underscore a growing emphasis on practical applications of AI in data science, particularly through enhanced coding agents and exploratory tools. The emergence of frameworks like the Model Context Protocol and new initiatives, such as the Agentic AI Foundation, aim to foster collaboration and transparency within the AI community. As organizations continue to develop sophisticated models with emergent capabilities and efficient tool-calling functions, the implications revolve around the need for ethics, effective use cases, and readiness to navigate the intricate dynamics of AI technology.