Glue on Pizza

From Curiosity to Practical Application

Will’s journey with AI began with a curiosity sparked by a Stanford online course on machine learning and neural networks some years ago. However, it was the recent explosion of generative AI tools that truly brought its potential into focus. “When AlphaGo hit the scene and just beat the top player by a long shot… that really woke me up to this idea that, hey, things are changing,” Will recalls, referencing the AI that mastered the complex game of Go.

AI in the Trenches: Infrastructure and Automation

While Will emphasizes that AI is not yet making autonomous decisions in their core infrastructure, it’s becoming an invaluable tool for:

1. Problem Solving and Troubleshooting: When faced with complex technical issues, Will and his team are increasingly turning to AI. “I took… a bunch of incomplete thoughts… a big list of, you know, this person did this… and I put all of that into… at the time we didn’t have any AI tools available [internally, so he used an external one]… and said, tell me, you know, come up with a creative performance appraisal.” While he ultimately wrote the appraisal himself, the AI helped organize thoughts, identify key points, and even suggest phrasing.
2. Coding and Scripting: AI is proving useful for generating and updating scripts, particularly in PowerShell. “Within the course of an afternoon, [I] was able to come up with a PowerShell script that does exactly what [I] needed to do, sends out an email report every Monday.”
3. Learning and Upskilling: For tasks that require new or less familiar coding languages or techniques, AI acts as a powerful learning aid and a way to quickly get up to speed.
4. Streamlining Repetitive Tasks: The potential to automate or semi-automate “busy work” is a significant draw, allowing the team to focus on more strategic initiatives.

The Human Element: Oversight and Critical Thinking

A recurring theme in Will’s approach is the critical role of human oversight:

• AI as a Tool, Not a Replacement: “I don’t want it to be the replacement for a TA, I don’t want it to be a replacement for office hours,” Will states, drawing parallels to the educational context. The goal is augmentation, not complete automation of human roles.
• Verifying AI Output: He stresses the importance of not blindly trusting AI-generated information or code. “It’s still just a glorified text prediction engine,” he cautions. Human expertise is still needed to validate, refine, and ensure the AI’s output is accurate and appropriate.
• Understanding the “Why”: Will is keen on using AI not just to get an answer, but to understand the process. When using AI for coding, for example, he wants to see the logic and steps, not just the final script.

Ethical Considerations and the “Black Box”

The conversation touched on the broader societal implications:

• Concentration of Power: Will expressed concern that AI could further concentrate power in the hands of those who control the technology.
• Job Displacement vs. Transformation: Like previous guests, he sees AI more as transforming tasks within jobs rather than causing widespread job elimination, but acknowledges the real anxieties around this.
• Bias and “Hallucinations”: The fact that AI models are trained on vast datasets means they can inherit and perpetuate biases present in that data. They can also “hallucinate” or generate incorrect information with conviction. “I think the only reason that you might come away with that opinion is because that just feels like such a… big part of this that we need to talk about,” he says, emphasizing the need for open discussion about these limitations.
• The “Opaque” Nature of AI: He highlighted the challenge of understanding how complex AI models arrive at their decisions, which can make it difficult to debug or fully trust their outputs in critical applications.

Wake Forest’s Approach: Deliberate and Thoughtful

Will appreciates Wake Forest’s approach to AI adoption, which he sees as deliberate and focused on responsible use. The university is encouraging exploration and learning, providing resources, and fostering conversations about both the potential and the pitfalls of AI.

Key Takeaway:

In the realm of IT infrastructure, AI is already proving to be a powerful tool for problem-solving, automation, and learning. Will Tomlinson’s experience underscores that while the technology is advancing rapidly, the human element – critical thinking, ethical oversight, and a clear understanding of AI’s limitations – remains paramount. The journey with AI is one of continuous learning and adaptation, and by embracing it thoughtfully, we can harness its power to enhance, rather than replace, human expertise.