DAILY BRIEF What changed in AI, dated and sourced. HUMANRANK.AI
HumanRank.ai Search models, tools, jobs⌘K Daily Brief Subscribe
BRIEF What changed in AI, dated and sourced.
HumanRank Brief

How to learn AI

Pick the path that matches your goal, then one course and a project of your own.

HUMANRANK EDITORIAL DESK · CHECKED 1 NOVEMBER 2026
Add as a preferred source on Google
Summarise with AI

How long the first AI course takes on each pathFIRST COURSE PER PATHHow long the first AI course takes on each pathUnder 5 hoursGoogle AI Essentials, touse AI at work1.5 hoursClaude Platform 101, tobuild with AI APIs, free3 hoursKaggle Intro to MachineLearning, no costUSD 49A month for Google AIEssentials after a 7-dayfree trialLengths and prices as listed on Google, Coursera, Anthropic and Kaggle course pages, checked 1 November 2026.How long the first AI course takes on each pathFIRST COURSE PER PATHHow long the first AI coursetakes on each pathUnder 5 hoursGoogle AI Essentials,to use AI at work1.5 hoursClaude Platform 101,to build with AI APIs,free3 hoursKaggle Intro toMachine Learning, nocostUSD 49A month for Google AIEssentials after a7-day free trialLengths and prices as listed on Google, Coursera,Anthropic and Kaggle course pages, checked 1November 2026.

For AI at work, start with Google AI Essentials (under 5 hours, USD 49 a month on Coursera after a 7-day trial); for building with APIs, the free Claude Platform 101; for ML engineering, Kaggle's free Intro to Machine Learning.

GOAL START WITH LENGTH PRICE (AS LISTED 18 OCTOBER 2026) THEN
Use AI at work Google AI Essentials 5 modules, under 5 hours USD 49 a month on Coursera after a 7-day free trial (US and Canada) Claude Academy AI Fluency, 4 hours, free
Build with AI APIs Claude Platform 101 13 lessons, 1.5 hours Free Building with the Claude API, 9 hours, free; OpenAI Academy Build with AI, free
Become an ML engineer Kaggle Intro to Machine Learning 3 hours No cost Machine Learning Specialization, 2 months at 10 hours a week; fast.ai, 9 lessons of about 90 minutes, free

Each path ends in a job family we cover: AI workflow design and prompt engineering for people who use AI at work, AI product management and AI evaluation for builders, and machine learning engineering for the technical path. AI ethics and governance roles draw on all three.

AI courses compared

Checked 18 October 2026 at each course's own page.
PLATFORMCOURSEWHO RUNS ITLENGTHPRICE (AS LISTED 18 OCTOBER 2026)BEST FOR
Google AI EssentialsAI Essentials certificateGoogle, on Coursera5 modules, under 5 hoursUSD 49 a month after a 7-day free trial in the US and Canada; prices differ by countryA first course for any office job
Claude AcademyAI Fluency: Framework and foundationsAnthropic14 lessons, 1 quiz, 4 hoursFreeLearning to delegate and check AI work
OpenAI AcademyApply AI at Work and Build with AI tracksOpenAISelf-paced; no length published on the course cardsFree; sign-up requiredChatGPT and Codex users
Elements of AIIntroduction to AI, then Building AIUniversity of Helsinki and MinnaLearn4 to 8 hours per part of Introduction to AI; about 50 hours for Building AIFreeA non-technical grounding
DeepLearning.AIAI For EveryoneDeepLearning.AI, on Coursera4 modules, 7 hoursFree to audit; a certificate needs a paid Coursera enrollmentManagers and non-engineers
Claude AcademyClaude Platform 101, Building with the Claude APIAnthropic1.5 hours; 9 hours (67 lessons, 8 quizzes)FreeDevelopers building on the Claude API
Microsoft LearnGet started with AI applications and agents on AzureMicrosoft7 modules, 5 hours 37 minutesFreeDevelopers who already know some Python
Hugging FaceLLM CourseHugging Face12 chapters plus setup, with a certification examFreeOpen-weight models and fine-tuning
KaggleIntro to Machine LearningKaggle3 hoursNo costYour first model, in a browser
DeepLearning.AIMachine Learning SpecializationDeepLearning.AI, on Coursera3 courses, 2 months at 10 hours a weekFree to audit; a certificate needs a paid Coursera enrollmentThe standard ML foundation
DeepLearning.AIDeep Learning SpecializationDeepLearning.AI, on Coursera5 courses, 3 months at 10 hours a weekFree to audit; a certificate needs a paid Coursera enrollmentNeural networks after the ML foundation
fast.aiPractical Deep Learning for Codersfast.ai9 lessons of about 90 minutes; part 2 adds more than 30 hoursFreeCoders who want to train real models fast

Verdicts in brief. Google AI Essentials is the fastest start for someone who will use AI tools at work and wants a certificate at the end. Claude Platform 101 and the OpenAI Academy courses are the quickest route to a first API call, and both cost nothing. Kaggle's Intro to Machine Learning is the lowest-friction first step toward engineering: no setup, no cost, three hours.

Coursera prices vary by country. On 18 October 2026 its Coursera Plus page showed us CA$81 a month or CA$552 a year, while Google quotes USD 49 a month for its certificates in the US and Canada. Every Coursera course above can be audited free, which means watching the videos and reading the material without graded work or a certificate.

What each path involves

Three goals, three different kinds of work:

Use AI at workYou learn to brief an assistant, judge its output and keep private data out of it. Claude Academy's AI Fluency course teaches a four-part method it calls Delegation, Description, Discernment and Diligence. Google AI Essentials runs the same ideas through workplace tasks: brainstorming, research and email.
Build with AI APIsYou learn to call a model from code, give it tools, retrieve documents and test the result. Claude Academy's Building with the Claude API covers prompting, tool use, RAG, agents and MCP in 67 lessons. Microsoft's Azure learning path asks for basic Python before you start.
Become a machine learning engineerYou learn the math and code behind models: regression, neural networks, training and evaluation. The Machine Learning Specialization builds models with NumPy, scikit-learn and TensorFlow; the Deep Learning Specialization is rated intermediate and takes about 3 months at 10 hours a week.
Prompting, whichever pathEvery path includes writing clear instructions. Our guide to writing better prompts covers the method the course providers teach, and our model profiles list the context limits and prices you will meet in the API courses.

The job titles differ too. People who use AI at work move into AI workflow design roles; builders become AI product managers or work in AI evaluation; the engineering path leads to machine learning engineer jobs, whose pay we track on our AI engineer salary page.

How to start

  1. 01DAY 1Name the goal in one sentence"Draft and check reports faster," "add a model to our app," or "get an ML engineering job." The goal picks the path; the path picks the course.
  2. 02WEEK 1Take one short courseUse AI at work: Google AI Essentials or AI Fluency. Build: Claude Platform 101 or an OpenAI Academy Build with AI course. Engineer: Kaggle's Intro to Machine Learning. All three take under 5 hours.
  3. 03WEEKS 2 TO 4Do one project of your ownA workflow you run every week, a small app that calls an API, or a model trained on a public dataset. Courses teach the vocabulary; a project proves you can use it.
  4. 04MONTHS 2 TO 3Go deeper only where the project got stuckBuilders move to Building with the Claude API or the Microsoft Azure path. Engineers move to the Machine Learning Specialization, then fast.ai or the Deep Learning Specialization.
  5. 05MONTH 3Put the work where employers lookA short write-up, a public repository or a demo. Our guide to becoming an AI engineer covers portfolios, and our entry level AI jobs guide lists paid evaluation work you can do while you learn.
Times are editorial estimates.

Before you sign up for an AI course

Before you sign up for an AI course
  • The course page names the organization that runs it and the date it was last updated.
  • You can see the syllabus, the length and the price before you create an account.
  • You know what the fee buys. On Coursera it buys graded work and a certificate; the videos are free to audit.
  • You never pay to apply for a job or a fellowship. A course fee is a fee for teaching, not a route to a guaranteed role.
  • The course uses a current model or library. Course pages that name a model version tell you how recent the material is.
  • You have a project in mind before you start, so you finish.

What the free material covers

The free routes are not watered down. Claude Academy's catalog, Microsoft Learn, Kaggle Learn, the Hugging Face course and fast.ai are complete courses with exercises; Hugging Face's LLM Course says it is "completely free and without ads" and ends in a certification exam. Elements of AI, built by the University of Helsinki and MinnaLearn, says more than 2 million people have taken its courses.

The API documentation is free too. The Claude Platform track links the API reference and cookbooks; OpenAI's and Google's developer docs do the same. Running the examples costs API usage, which starts low: our model profiles list GPT-6 Luna at USD 0.10 per 1M input tokens and Gemini 3.5 Flash-Lite at USD 0.30, as checked in October 2026.

What you pay for is structure and proof: a graded path, a deadline and a certificate you can show. Elements of AI recommends finishing within six weeks because, in its experience, a deadline makes students more likely to finish.

Sources

  1. [1]Google AI Essentials, Google, course page and FAQ
  2. [2]Claude Academy, Anthropic, course catalog and Claude Platform track
  3. [3]OpenAI Academy courses, OpenAI
  4. [4]Get started with AI applications and agents on Azure, Microsoft Learn
  5. [5]Practical Deep Learning for Coders, fast.ai
  6. [6]Intro to Machine Learning, Kaggle Learn
  7. [7]LLM Course, Hugging Face
  8. [8]Elements of AI and its course length FAQ, University of Helsinki and MinnaLearn
  9. [9]AI For Everyone, Machine Learning Specialization and Deep Learning Specialization, DeepLearning.AI on Coursera
  10. [10]Coursera Plus, Coursera

How we researched this

On 18 October 2026 we opened each course's own page and recorded who runs it, its length as published, its price or free status, and its prerequisites. Where a provider publishes no length (OpenAI Academy course cards) or prices by country (Coursera), we say so instead of estimating. We grouped courses by the goal they serve and matched each goal to the career paths HumanRank already covers.

Frequently asked questions

How do I learn AI from scratch with no coding?

Start with Google AI Essentials (under 5 hours) or Elements of AI's Introduction to AI (4 to 8 hours per part, free). Neither requires programming. Add Claude Academy's AI Fluency course for a method you can apply to any assistant.

Which AI courses are free?

Claude Academy, OpenAI Academy, Microsoft Learn, Kaggle Learn, the Hugging Face LLM Course, fast.ai and Elements of AI are free. DeepLearning.AI's Coursera courses are free to audit; the certificate is paid.

How long does it take to learn AI?

A working grounding takes under 10 hours on any of the starter courses. The engineering path takes months: the Machine Learning Specialization lists 2 months at 10 hours a week and the Deep Learning Specialization 3 months at the same pace, before you build a portfolio.

Do I need math to learn AI?

Not for the first two paths. For the engineering path, fast.ai says you need no university math because it teaches the calculus and linear algebra you need during the course; the Machine Learning Specialization is rated beginner.

Is a Google AI Essentials certificate worth it?

It costs USD 49 a month in the US and Canada and takes under 5 hours, so most people finish inside one billing month. It shows an employer you can use AI tools; it does not qualify you for engineering roles.

Which path leads to a job fastest?

Paid evaluation and annotation work is open now and needs no course; see our entry level AI jobs guide. Salaried engineering roles need the longer path plus a portfolio, covered in how to become an AI engineer.

Should I learn with the Claude API or the OpenAI API?

Both publish free courses and documentation, and the concepts transfer. Pick the one whose models you plan to ship with; our model profiles list prices and limits side by side.