AI in technical interviews: The future of hiring or a major risk?
James Wilson
Specialist in AI & ML Software Engineers through to Engineering Leaders
If we are connected on LinkedIn, you might have seen my recent post about technical interviews and the suspicious use of dual monitors—or as I like to call it, the “AI paranoia”. It went viral (link here, in case you missed the drama), and boy, did it spark a conversation! It seems like everyone has an opinion about how tech interviews should be conducted, and honestly, I loved every bit of the discussion that followed.
So, why not dive a bit deeper in this newsletter edition? I think the post struck a nerve because it highlighted a growing disconnect: traditional interview norms versus the everyday realities of tech work, where AI and other tools are practically non-negotiables for boosting productivity. But the real conversation goes beyond just AI; it's about what we’re actually testing for in interviews versus what truly matters on the job. It’s about reconciling what we expect candidates to showcase in a high-stakes setting and how they genuinely operate in their roles every day.
Job interviews are supposed to be a sneak peek into how someone actually works, right? Yet, too often, they end up feeling like a memory contest. In the real world, nobody codes from memory or avoids Google because it’s considered “cheating.” Engineers consult docs, bounce ideas off colleagues, and, of course, use AI. So why should interviews be any different?
Here’s where my slightly controversial take comes in: sometimes, it makes sense for hiring managers to design a technical interview that doesn’t exactly mirror the day-to-day. Why? Because within the tight timeframe of an interview, it’s tough to truly gauge how a candidate thinks, collaborates, and adapts, or to test for specific knowledge that’s crucial for hitting the ground running. In some cases, an unassisted interview—where candidates rely solely on their own knowledge and reasoning—can provide a clearer snapshot of their problem-solving approach without the safety net of Google or AI. This strategy can be a valuable way to see how they’d handle complex, novel challenges that demand deep thinking and innovation. After all, while tools can boost productivity, the real value lies in a candidate’s ability to think critically and creatively, especially when the going gets tough.
Do I think that all technical interviews should be unassisted? Certainly not! The reality is that most of the work in tech involves using resources effectively. A great engineer isn’t necessarily the one who can remember every syntax rule off the top of their head, but the one who knows how to quickly find the right answers and apply them efficiently.?
Ultimately, the key is balance. Some roles may benefit from seeing a candidate’s raw thought process in an unassisted setting. For others, understanding how they use their resources to get the job done is far more relevant. It’s about tailoring the interview approach to the needs of the role and the realities of the job, not rigidly sticking to one format. Flexibility in the interview design can lead to a richer understanding of a candidate’s potential, allowing smarter hiring decisions that align with the actual work environment.
领英推荐
Regardless of the interview approach —assisted or unassisted—one thing remains non-negotiable: integrity. The goal is to create a fair playing field where candidates can showcase their skills honestly and transparently. That’s why it is crucial for hiring managers to set crystal-clear expectations from the outset. If you are an interviewer running an unassisted interview, make it known upfront: no AI, no Google, no extra monitors. If you’re allowing candidates to use resources, specify exactly what’s acceptable. Clear expectations help candidates prepare appropriately and prevent any misunderstandings or accidental?rule-breaking.
Equally important is explaining the rationale behind your chosen approach. Whether you opt for assisted or unassisted interviews, candidates deserve to understand the “why” behind the rules. If you are going unassisted to gauge deep problem-solving skills or raw thinking, say so. If you are allowing resources to see how well they can leverage tools to get the job done, explain that too. Being transparent about your reasoning not only sets the stage for a fair interview but also demonstrates respect for the candidate’s time and effort. It shows that the process is thoughtful and purposeful—not just a checkbox exercise.
And let’s be real: the strong response to my original post shows that certain interview practices can alienate candidates. If an interview feels disconnected from the actual work environment or sets unclear expectations, it risks turning away great talent who might excel under different circumstances. At the end of the day, the goal is to attract and assess the best fit for the role, not to trip candidates up with unexpected twists. By being upfront about your interview format and its purpose, you create a more inclusive, respectful process that aligns with the realities of the job—and that’s (hopefully) something everyone can get behind.
Interested in more insights like these? Please subscribe to my newsletter to receive notifications.
Education Management Professional
1 个月Yes and what truly matters on the job will be understood by someone who has done the job over many years and appreciates the nuances and complexities one may meet. You can’t build nuance into AI or relevant real life experience into AI Msybe the companies who wish to go down this route want to install as many machines as possible. Cause Machines don’t need , food breaks, toilet breaks, complain or be part of a union, Don’t need workmplace pensions , Don’t need annual leave. Most not all of this AI boils down to one very human trait. Greed . Greed on part of big tech and greed on part of big business that wants to avoid the store mentioned costs. Dress it up how you want you ain’t fooling me!
Education Management Professional
1 个月A major risk. A human needs to decide on the best person for the job not AI algorithms. If an algorithm gets the wrong person then who is accountable.Ans no one because the whole AI input is so vast with varied contributions all accountability is removed. Welcome to the new dystopian worl our tech profs are producing.The giant tech companies only care about wonga. They will keep pushing for more and more uses of AI regardless of the foreseen consequences !
Data Management | Data Analytics | Data Visualization | Data Governance | Automation | Software Engineering
1 个月A few years back, I walked out of an unassisted interview because the question made no sense to me- the question was about stream handling for a job in C# and all he really wanted was the .NET object hierarchy to get there, and I went into a diatribe on how C became C++ became Java became C# and the full history behind it. I got frustrated with him repeating the question and walked out and only on the way home did I realize what he was asking.. I put that out there because the current generation of AI is like an autistic 13 year old girl- quite certain, but doesn't even know 1/10th of what a 30, 40, or 50 year programmer knows. This is good for evaluating a recent college graduate, and transparency is good for anybody and everybody- but the best jobs I ever had were ones where the technical interview came from a walkthrough of the actual project design and not specifically testing for possibly irrelevant skills.
All things data, investments, and leadership.
1 个月Great points and very pragmatic. If I may: I'd suggest that baking in the decision-making and rules for how to handle applicants with #neurodiversity accomodation needs should be done up-front as part of this process. If you wait until after it comes up, you'll get some friction in terms of deviating from standards and the statistically predictable resistance predicated on "unfair advantage" or "two tier process". If you are thoughtful about your full process for all candidates and outline it all from scratch, you can bake in a lot of the thinking on how you deal with "accommodation requests".