Stated Hours vs Real Weeks: How Long Popular AI Programs Take

How we turned hours into weeks

We convert stated hours into derived weeks by dividing listed hours by 3, 5, and 10 hours/week to one decimal: on the Machine Learning Specialization (as shown in the United States on 2 November 2026), "2 months at 10 hours a week" implies 80 to 87 derived hours.

Weeks equal hours/pace to one decimal (Machine Learning at One Hour a Day), with undefined months at 4 and 4.35 weeks (our assumption). No page captured on 2 November 2026 states a measured completion time or rate; grow.google (Google Career Essentials are self-paced; many finish within one month, 2 November 2026) and DeepLearning.AI's membership FAQ (most spend 3-5 hours/week on coursework, 2 November 2026) give provider wording with no measurement behind it (how to choose an AI course).

Short programmes: when the same course has three numbers

A short self-paced AI programme can state several lengths: Google AI Essentials shows "<5 hours to complete" on grow.google (2 November 2026), "4 hours to complete" and "In under 10 hours" on Coursera (2 November 2026), and course lines totaling 8 hours (1.6 derived weeks at 5 hours/week), as shown in the United States.

At 3/5/10h/wk (derived): Google's self-paced Google AI Essentials on grow.google (2026-11-02) shows "<5 hours to complete" and "In under 5 hours" (1.7/1.0/0.5w upper limits); Coursera (2026-11-02) shows "4 hours to complete" (1.3/0.8/0.4w), "In under 10 hours", and 5 courses (1, 2, 2, 1, 2h) totaling 8h (2.7/1.6/0.8w). Self-paced Google AI Professional Certificate (2026-11-02) shows "8 hours to complete" (2.7/1.6/0.8w) and 8 FAQ courses at ~1h each, vs 8 courses (3, 1, 1, 1, 2, 1, 2, 2h) totaling 13h (4.3/2.6/1.3w). DeepLearning.AI's self-paced Agentic AI (2026-11-02) shows 9h55m (9.9h: 3.3/2.0/1.0w) with no pace line.

Coursera sequences: headline, FAQ and course lines

Multi-course Coursera sequences can differ across headline, FAQ, and course-line totals: IBM AI Developer lists ten courses totaling 146 hours, or 29.2 derived weeks at 5 hours/week, against a "6 months at 4 hours a week" headline (19.2 to 20.9 derived weeks), as shown in the United States on 2 November 2026.

At 3/5/10h/wk (derived, 2026-11-02): Machine Learning Specialization shows 94h58m on DeepLearning.AI and 3 courses (33, 34, 28h) on Coursera totaling 95h (31.7/19.0/9.5w), vs "2 months at 10 hours a week" (80 to 87h: 26.7 to 29.0, 16.0 to 17.4, 8.0 to 8.7w) and both FAQs' 3, 4, 3w at 5h/wk (50h: 16.7/10.0/5.0w). Deep Learning Specialization lists 5 courses (25, 24, 7, 36, 37h) totaling 129h (43.0/25.8/12.9w), vs "3 months at 10 hours a week" (120 to 130.5h: 40.0 to 43.5, 24.0 to 26.1, 12.0 to 13.1w) and 5 FAQ courses at 5w each at 5h/wk except course 3 at ~4w (24.0w, 120h: 40.0/24.0/12.0w).

At 3/5/10h/wk (derived), on IBM's IBM AI Developer Professional Certificate (2026-11-02), the 6-month/4h headline is 96 to 104.4h (32.0 to 34.8, 19.2 to 20.9, 9.6 to 10.4w), the FAQ gives 4-6 weeks per self-paced course at 3-5 hours/week (6-8 months), and 10 courses (15, 13, 8, 10, 15, 24, 12, 15, 23, 11h) total 146h (48.7/29.2/14.6w). On 2026-11-02, IBM AI Enterprise Workflow Specialization shows "2 months to complete at 10 hours a week" (80 to 87h: 26.7 to 29.0, 16.0 to 17.4, 8.0 to 8.7w), an FAQ of 35-40 hours across 6 courses of 4 to 9 hours (11.7 to 13.3, 7.0 to 8.0, 3.5 to 4.0w), and 6 courses (8, 11, 12, 14, 9, 17h) totaling 71h (23.7/14.2/7.1w); Microsoft's Microsoft AI & ML Engineering Professional Certificate shows "6 months at 7 hours a week" (168 to 182.7h: 56.0 to 60.9, 33.6 to 36.5, 16.8 to 18.3w), an FAQ of approximately 35-40 hours per course across 5 courses (175 to 200h: 58.3 to 66.7, 35.0 to 40.0, 17.5 to 20.0w), and 5 courses (36, 45, 43, 21, 31h) totaling 176h (58.7/35.2/17.6w).

CS50 AI: no total, and two weekly ranges

Harvard's CS50 AI states no lecture total on its course or edX pages, while edX lists "7 weeks" at "10-30 hours per week" (as shown in the United States on 2 November 2026), giving 70 to 210 hours or 14.0 to 42.0 derived weeks at 5 hours/week.

On edX, CS50's Introduction to Artificial Intelligence with Python (2026-11-02) shows "Self-paced", "7 weeks", "10-30 hours per week", and "Ends Jun 30, 2027" (70 to 210h: 23.3 to 70.0, 14.0 to 42.0, 7.0 to 21.0 derived weeks at 3/5/10h/wk), while CS50 OpenCourseWare (2026-11-02) offers seven weeks free. The CS50 FAQ (2026-11-02) says to add lecture times yourself and estimates 10-20 hours/week for problem sets (70 to 140h without lectures: 23.3 to 46.7, 14.0 to 28.0, 7.0 to 14.0 derived weeks), with no individual deadlines and a current final deadline of 30 June 2027, with no exceptions.

Elements of AI: hours per part, and a recommended deadline

Elements of AI states no total hours for Introduction to AI on its home or overview pages, while its FAQ estimates 4-8 hours/part (as shown in the United States on 2 November 2026): if one part is one chapter, which is our reading, that is 4.8 to 9.6 derived weeks at 5 hours/week.

From University of Helsinki and MinnaLearn, the Elements of AI: Introduction to AI home page describes free self-paced courses and the course overview lists Chapter 1 to Chapter 6 as free, stating no hours (2026-11-02). The FAQ (2026-11-02) estimates 4-8 hours/part (more with links) and recommends 6 weeks; if one part is one chapter, which is our reading, 24 to 48 hours implies 4 to 8h/wk over 6 weeks (8.0 to 16.0, 4.8 to 9.6, 2.4 to 4.8 derived weeks at 3/5/10h/wk), while Building AI estimates ~50 hours depending on difficulty and the optional project (16.7/10.0/5.0 derived weeks).

Access windows: Stanford and HBS Online

Fixed 90-day enrollment windows set a floor on weekly study pace: Stanford Online's Agentic AI Program states 30 hours to earn the certificate and 90 days of access (2 November 2026), or >=2.3 derived hours/week, while HBS Online's AI Essentials for Business states 16-24 hours and a 90-day window (2 November 2026), as shown in the United States.

On Stanford Online's self-paced, on-demand Stanford Agentic AI Program (2026-11-02), "90 days access, 30 hours to earn the certificate" covers ~30 hours of video and coding assignments (10.0/6.0/3.0 derived weeks at 3/5/10h/wk), requiring >=2.3h/wk across 90 days (12.9 derived weeks) with no extension stated. On Harvard Business School Online's AI Essentials for Business (2026-11-02), "Format On-Demand", "Start Anytime", "Time 16-24 hrs, 4-6 hrs/module", and "4 Weeks, 16-24 Hours, 90-day access window" imply 4 to 6h/wk over 4 weeks (5.3 to 8.0, 3.2 to 4.8, 1.6 to 2.4 derived weeks), while modules (5-6, 4-5, 5-6, 4-5h) sum to 18 to 22 derived hours; the 90-day window requires >=1.2 to 1.9 derived hours/week; see live cohort vs self-paced AI programs.

Weeks at 3, 5 and 10 hours, programme by programme

Across these 12 AI programmes observed in the United States on 2 November 2026, derived time at 5 hours/week spans from 1.6 weeks on Google AI Essentials (8 course-line hours, 2 November 2026) to 42.0 weeks at the top of CS50 AI's edX range (210 hours, 2 November 2026).

Course-line sums are the basis where listed; week columns are derived (hours/pace, not a ranking or learner measurement).

Hidden costs and failure points

Going slower than a programme's headline pace can trigger subscription renewals, trial expirations, or access cutoffs: Coursera subscriptions ($49/month on three programmes as shown in the United States on 2 November 2026) auto-renew until cancelled, and access runs 90 days at Stanford (2 November 2026) and HBS Online (2 November 2026).

See true cost of an AI certificate.

• Coursera: In the U.S./Canada, Google AI Essentials is $49 USD/mo after a 7-day trial on grow.google (2026-11-02) and $49/mo after a 7-day trial on Coursera (2026-11-02); on 2026-11-02, Google AI Certificate ($49/mo) and ML Specialization ($49 / month) are priced, Deep Learning's FAQ says it cannot be taken for free (financial aid can be applied for), and Deep Learning, IBM Workflow, Microsoft AI & ML, plus IBM Developer (2026-11-02) show no accepted price. Coursera Plus (separate plan, not a programme price) is $59/mo (cancel anytime, 7-day trial) or $399/yr (14-day money-back guarantee), 2026-11-02. Terms (2026-11-02) bar trial certificates, auto-renew for equal periods until cancelled, and allow Specialization refunds within 7 days or before a certificate, whichever is first; cancelling stops billing without automatic refund.
• DeepLearning.AI: Pro is $25/mo annual or $30/mo monthly on ML Specialization (2026-11-02) and Agentic AI (2026-11-02), in USD exclusive of taxes unless stated (2026-11-02); Agentic AI free audit gives videos/forum without assessments, certificates, or project tools (2026-11-02).
• Trials, windows, deadlines: Microsoft AI & ML needs an Azure license or time-limited free trial that may expire before completion (2026-11-02). Stanford costs $1,650 tuition plus ~$100 compute (90-day access, no extension stated, 2026-11-02). HBS Online costs $1,949 (90-day access/certificate window from enrollment, 2026-11-02). CS50 AI is free on OpenCourseWare with no accepted price for edX verification (2026-11-02) and a current final deadline of 30 June 2027, with no exceptions (2026-11-02).

FAQ

The four answers below cover how to calculate derived weeks at 5 hours/week, why figures on one course page disagree, why derived weeks are not measured completion times, and what limits apply, as shown in the United States on 2 November 2026, including CS50 AI's 30 June 2027 deadline.

How long will an AI course take at 5 hours a week?

Divide its listed hours by 5. By their own course lines, Google AI Professional Certificate comes to 2.6 weeks, Machine Learning Specialization to 19.0 weeks and Microsoft AI & ML Engineering to 35.2 weeks (derived, as shown in the United States on 2 November 2026).

Why do the numbers on one course page disagree?

These lines do not match. IBM AI Enterprise Workflow shows 2 months at 10 hours a week, 35-40 hours in its FAQ, and six course lines that add up to 71 hours (as shown in the United States on 2 November 2026). The page does not explain the difference.

Are these real completion times?

No. They are arithmetic on the provider's own lines, with weeks equal to hours divided by pace. None of the pages captured on 2 November 2026 reports a measured completion time or rate. DeepLearning.AI says most learners spend 3-5 hours per week on coursework, with no figures behind it (2 November 2026).

What happens if I go slower than the page assumes?

It depends on the limit. Coursera subscriptions renew until cancelled, Stanford's Agentic AI Program includes 90 days of access, and CS50 AI has a final deadline of 30 June 2027 (Terms and edX 2 November 2026; Stanford 2 November 2026; FAQ 2 November 2026).

Sources

The 22 official pages below support every stated hour figure, FAQ estimate, price state, and access rule in this guide, as shown in the United States on 2 November 2026 (22).

1. Google (2026-11-02)
2. Essentials (2026-11-02)
3. Certificate (2026-11-02)
4. DeepLearning (2026-11-02)
5. ML (2026-11-02)
6. Deep (2026-11-02)
7. Developer (2026-11-02)
8. Workflow (2026-11-02)
9. Microsoft (2026-11-02)
10. FAQs (2026-11-02)
11. CS50 (2026-11-02)
12. edX (2026-11-02)
13. Hours (2026-11-02)
14. Elements (2026-11-02)
15. Overview (2026-11-02)
16. Agentic (2026-11-02)
17. Membership (2026-11-02)
18. Cost (2026-11-02)
19. Stanford (2026-11-02)
20. HBS (2026-11-02)
21. Plus (2026-11-02)
22. Terms (2026-11-02)

Compare listed program details →

Last reviewed: 2 November 2026. See the editorial policy.

Program Change Monitor · every two weeks

Three programs whose price or content changed, plus one new decision guide. Sourced and dated, no marketing.

Email address Subscribe

Double opt-in: you confirm by email before anything is sent. Unsubscribe in one click in every issue. Your address is stored by Buttondown (USA) under a data-processing agreement. See our privacy notice.

← All decision guides