Oracle got AI working on itself only this year

Oracle Embraces AI Internally: A Rapid Transformation

Oracle has achieved a significant milestone this year by successfully integrating AI into its own operations, marking a dramatic shift in its approach to artificial intelligence.

A Surprising Turn of Events

  • Rapid Adoption: Within three months, 80% of Oracle's staff adopted the new AI tools, with developers completing tasks in a week that previously took two quarters.
  • Investment and Infrastructure: Over the past two years, Oracle has invested tens of billions of dollars to build the necessary infrastructure for AI, yet it wasn't until April that its employees gained access to a broadly useful AI tool.

Launching AI Tools

In April and May, Oracle introduced ChatGPT Enterprise and OpenAI's coding tool Codex, enabling AI integration across various departments, including customer support, developers, finance, and sales.

Implementing AI with Care

Jae Evans, Oracle's global chief information officer, highlights the company's meticulous approach to implementing AI:

  • Corporate Standards: Oracle established corporate standards, security controls, and internal policies before launching the tools.
  • Cost Management: The company now tracks the AI models used by employees and their associated costs, redirecting routine tasks to more cost-effective models like GPT-5.6 Terra.

Productivity Boosts and Remaining Challenges

  • Increased Productivity: Developers now complete tasks in a week that previously took two to three quarters, thanks to AI assistance.
  • Catching Up: However, Oracle is still updating its testing, validation, deployment, and release management processes to keep up with the rapid pace of AI-driven development.

Security Considerations

  • Anthropic's Claude Mythos Preview: Oracle was among the select partners granted early access to this powerful model, which identified and exploited zero-day vulnerabilities in major operating systems and browsers.
  • Verification Challenges: Oracle found a significant number of potential vulnerabilities in its code within two weeks, but a high false-positive rate (60% to 70%) required additional verification steps before fixes could be implemented.

Open Communication and Ongoing Adjustments

Oracle's transparency in discussing these challenges, alongside its ongoing job cuts and significant data center investments, sets a unique tone in the industry. The company's journey with AI is a testament to the power and complexities of this rapidly evolving technology.

Leave a Reply