How well can LLMs do in ML/DS interviews?
We have recently written a paper publishing interesting results about this very question. Check it out on Arxiv here.
In our experiments, it appears that models like OpenAI's o1 are on par with Senior Machine Learning Engineers but they aren't perfect! Our abstract summarizes it well:
We present HardML, a benchmark designed to evaluate the knowledge and reasoning abilities in the fields of data science and machine learning. HardML comprises a diverse set of 100 challenging multiple-choice questions, handcrafted over a period of 6 months, covering the most popular and modern branches of data science and machine learning. These questions are challenging even for a typical Senior Machine Learning Engineer to answer correctly. To minimize the risk of data contamination, HardML uses mostly original content devised by the author. Current state-of-the-art AI models achieve a 30% error rate on this benchmark, which is about 3 times larger than the one achieved on the equivalent, well-known MMLU ML. While HardML is limited in scope and not aiming to push the frontier, primarily due to its multiple-choice nature, it serves as a rigorous and modern testbed to quantify and track the progress of top AI. While plenty of benchmarks and experimentation in LLM evaluation exist in other STEM fields like mathematics, physics, and chemistry, the subfields of data science and machine learning remain fairly underexplored.
These findings indicate that while AI has made substantial progress in answering ML-related questions, there is still a gap between machine intelligence and expert human reasoning. This gap presents opportunities for further advancements in AI training, prompting research into methods that enhance reasoning and comprehension.
In future studies, we plan to expand HardML by incorporating open-ended questions and real-world coding challenges. These additions will provide a more comprehensive evaluation of AI's capabilities in tackling complex ML problems beyond multiple-choice questions.
Stay tuned for more insights as we continue to analyze the frontier of AI in ML interviews!
