Building Systems with the ChatGPT API vs AI for Medicine Specialization

Across 11 short lessons (roughly 90 minutes total), the course covers a complete pipeline for multi-step LLM systems: how language models and tokenisation work, the chat format and system-user message separation, input classification for query routing, the OpenAI Moderation API, chain-of-thought prompting to handle multi-step questions, chaining several focused prompts where each consumes the previous output, output checking, and a two-part section on evaluating LLM responses at the system level. Reviewers consistently praise the logical progression and the theory-to-practice balance. The principal mark-down is age and depth: the course was built on GPT-3.5 Turbo in 2023 and has not been meaningfully updated, so it predates tool calling, structured JSON outputs, and reasoning models, and it stops short of real-world deployment concerns such as latency management, cost at scale, and production observability.

Isa Fulford, Member of Technical Staff at OpenAI, leads the code demonstrations while Andrew Ng frames the broader concepts and asks the questions a beginner would actually ask. Reviewers across blogs and Coursera call the pairing "highly knowledgeable and effective communicators." The teacher-demonstrator dynamic mirrors how a learner thinks through a new problem step by step, keeping each lesson of five to twenty minutes focused and coherent. Because Fulford comes directly from the team that built the ChatGPT API, the design decisions behind the Moderation API, the chat format, and tokenisation carry genuine authority rather than third-hand explanation.

The course is free on the DeepLearning.AI platform with every Jupyter notebook runnable directly in-browser — no OpenAI API key, no local Python environment, and no subscription required. The Coursera guided-project version is also free to audit. For roughly 90 minutes of hands-on instruction from two of the most credible names in the field, delivering reusable architecture patterns for multi-step LLM systems, the value proposition is essentially unmatched among paid or free alternatives. The only caveats are that a graded assignment and certificate on the Coursera version sit behind a paid enrolment, and the free tier leaves no portfolio artefact by default.

The patterns taught — classify the input, moderate for safety, reason in steps, chain focused prompts rather than one monolithic prompt, then evaluate the output — are exactly how production LLM features are structured in practice. Multiple reviewers note that the progression from basic API calls to a multi-stage orchestrated system reflects real engineering work. The gap is that the 2023 course predates the patterns now central to production LLM development (tool calling, structured outputs, retrieval-augmented generation), and at least one practitioner reviewer noted that the finished chatbot example would require substantial hardening before it approached something ready for deployment beyond a prototype.

Every lesson pairs a video with a runnable Jupyter notebook, and the course builds one coherent end-to-end example: a customer-service chatbot that classifies incoming queries, runs them through the Moderation API, applies chain-of-thought prompting to multi-step reasoning, chains successive focused prompts, retrieves product information, and evaluates whether its own output actually addresses the user's question. The Coursera version holds a 4.7/5 rating across 346 learners. The caveat is that there is no graded project or kept portfolio artefact on the free tier, and the supplied notebooks now require fixes (deprecated API syntax, missing helper files) to run locally outside the course sandbox.

The specialization covers an unusually well-chosen slice of applied medical AI: CNN classification and U-Net segmentation on chest X-rays and 3D brain MRIs (Course 1), tree-based risk models, random forests, and survival/hazard estimators (Course 2), and causal treatment-effect estimation, GradCAM/SHAP/permutation-importance interpretation, plus BERT-based NLP label extraction from radiology reports (Course 3). Coursera learners describe "extremely well-written content/code and short but illuminating lectures" and "good terse discussions of common metrics, issues with imbalanced datasets... U-Net architecture and loss functions for semantic segmentation." The recurring content criticism is depth: reviewers note "very terse explanation of ROC curve," that the specialization "misses in depth theory," and that "many things were abstracted away," leaving some unsure they could replicate the methods unaided. It teaches application patterns excellently but is not a from-scratch theory course.

Lead instructor Pranav Rajpurkar — a Stanford researcher and lead author of the landmark CheXNet paper that first matched radiologists at detecting pneumonia from chest X-rays — is the most consistently praised element of the program, supported by co-instructors Bora Uyumazturk, Amirhossein Kiani, and Eddy Shyu. Coursera learners call him "extremely thorough" and say "by employing intuitive figures and examples in his presentations, he makes even the most nuanced topics easy to follow." The instructor rating sits at 4.7/5. The only consistent reservation is delivery pacing — videos are short and dense, which some learners want expanded for harder concepts like survival analysis and causal inference.

The specialization is delivered on a subscription basis: roughly $49/month on Coursera (or about $30/month via a DeepLearning.AI Pro subscription), with the entire first module previewable for free. Because a motivated learner can finish all three courses in roughly 9–12 weeks at 4–6 hours per week, the total cash outlay is typically one to three monthly payments — modest for the specialized, hard-to-find medical-AI content and the named Stanford instruction. Reviewers on Shiksha and Class Central treat it as good value for the niche, though the value proposition weakens for learners who lack the deep-learning prerequisites and end up paying additional months while they backfill foundations from the (separate) Deep Learning Specialization.

As a self-paced MOOC, direct support is limited to discussion forums and peer interaction rather than instructor contact, which is standard for Coursera specializations. The most concrete support-related friction reported by learners is the auto-grader: multiple reviewers "knocked down a star rating for the finicky auto-grader" and wished it would "provide more instructive feedback than just correct/incorrect," with specific complaints about completing the Week 3 programming assignment. Several also note the notebooks run only inside the Coursera environment ("the codes do not work in Google Colab"), so learners who hit environment issues have limited recourse beyond the forums.

This is the specialization's strongest differentiator. Rather than toy datasets, learners work with realistic medical imaging, survival data, and clinical text, and learn the practical nuances practitioners actually face — class imbalance, patient overlap between train/test splits, evaluation with sensitivity/specificity and ROC, censored survival data, randomized-trial treatment effects, and explainability methods clinicians demand. A learner from a medical-imaging background wrote "I can't express how useful and precise were your teaching materials," and the program is repeatedly recommended for professionals with some ML background who want to move into the healthcare-AI space. The caveat is that production deployment, regulatory, and data-engineering realities of real clinical systems are outside scope.