← Back

Build Multimodal Generative AI Applications

IBM · View on Coursera ↗

Or see the score while you browse.

Taught by Hailey Quach, Ricky Shi

Computer ScienceData Science

Coursera
4.6
headline rating
↓
1.4
gap
Starisk
3.2
Padded
Small sample. Only 11 reviews were readable, so this score is a hint, not a verdict. Every other course page reads about 20 of the most-helpful reviews.

About the course

Ready to level up your GenAI skills? Step into the exciting world of multimodal AI, where language, images, and speech come together to build smarter, more interactive applications. In this hands-on course, you’ll learn how to build systems that work across multiple modalities, from creating AI-powered storytellers and meeting assistants to developing image captioning tools and video generation apps.

Read more

You’ll gain experience with real-world tools like IBM’s Granite, OpenAI’s Whisper, Sora and DALL·E, Meta’s Llama, Mistral’s Mixtral, and Gradio. Plus, you'll explore multimodal search, question answering, and retrieval systems that combine text, speech, and visual data. By the end of the course, you’ll be able to design and build full-stack multimodal AI solutions using Python and frameworks like Flask and Gradio. If you’re looking to gain in-demand skills for building the next generation of AI applications, enroll today and power up your AI career!

The dissent

The most-upvoted critical review (1 vote, 1.0★).

“I gave 2 stars for the courses before this gets one star. I am two far progressed to stop now, but the whole specialization is really bad from my point of view. There is hardly any structure, its more topic based mostly random information and questions. The questions could also be answered with common sence. The whole courses could also be boilt down to 2 or 3 courses, but in this way it looks like you achieved more I guess..”

1.0 / 5

Read the full review on Coursera ↗

Coursera's 4.6 comes from 66 ratings. Of those, 6.1% are 1–2★.

Starisk reads the reviews learners voted most helpful, and weights each by the square root of its votes: a review with 80 upvotes counts about five times one with 2. Unvoted reviews are not skipped out of laziness — by this model they carry no evidence that other students endorsed them.

That measures what the most-endorsed reviews say, not “true quality.” The rating distribution above is the whole population, so you can see the sample next to it.

Reviews read on 3 Nov 2026.Read them on Coursera ↗