Richard Johnson

An opinionated single-pick recommendation of the best AI research tool for students, why Consensus wins for most, the best free-only option, and when to pick something else.

What the main AI research tools cost for students: the free tiers, the paid plans, and which ones offer a genuine student or academic discount.

A side-by-side comparison of the main AI research tools, Consensus, Elicit, Scite, Perplexity, Semantic Scholar, and Research Rabbit, on what each does best and where it falls short.

The three different limits on AI tools explained: usage caps, API rate limits, and the context window, why they exist, and how to stop hitting them.

Why AI models get engineering unit conversions wrong, which conversions invite which mistakes, and a five-step workflow to keep the error out of your work.

A repeatable protocol for testing whether an AI tool is accurate enough for engineering calculations, paired with what independent published benchmarks actually found.

A step-by-step cable sizing workflow with an AI assistant in the loop: gather inputs, check corrected ampacity, voltage drop, and short-circuit withstand, then confirm every number against the NEC or IEC table.

A side-by-side price comparison of the main AI tools engineers choose between: free, individual (around $20/mo), power ($100 to $200/mo), team per-seat, and quote-based enterprise tiers.

Which AI tools and tiers do not train on your inputs by default: the business and API tiers of OpenAI, Anthropic, Google, and Microsoft, versus the consumer tiers that train unless you opt out.

What structural engineers say about AI tools, synthesized honestly: used for drafting and scripting, kept off the analysis and load path, with a reality check on the tools.

Stop letting a chat AI produce your engineering numbers: it predicts tokens, not computes. Use AI to set up and explain, a computational engine to solve, and verify.