The Current Dearth of AI Certifications Seems Likely to Continue

In May, I wrote a column for GoCertify about the "ChatGPT Gold Rush." This morning, I just got through a TechTarget WhatIs.com story titled "10 Top Artificial Intelligence Certifications and Courses for 2023." (It’s by an old colleague and familiar writer, Andy Patrizio, and definitely worth a read.)

‍

A common thread is emerging and it's probably not one that readers will like a whole lot. As I found when looking at ChatGPT in particular, and what Mr. P. found when looking more broadly at the whole field of AI/ML can best be summed up as follows:

‍

1) There are many, many more courses, including certificate and degree programs (most at the graduate level), than there are certifications available on AI topics right now. On the plus side, the TechTarget piece cites some big-name, high-value institutions — namely, Stanford, MIT, and UC Berkeley. That three of 10 items are from high-dollar universities also speaks to certain cost and time considerations that factor into taking on the subject.

2) Two major training companies are involved in more than half of the 10 items covered, namely Coursera (3; items 4-6) and Udemy (2; items 7 and 8). Both of these players offer good- to high-quality courseware and content, but are famous for getting stuff out the door, fast. Ditto for LinkedIn Learning which is also represented on the list (item 10).

3) Right now, IBM is the only cert sponsor with a vendor-specific offering in this list — it's based on IBM Watson AI Services, APIs and Python, plus OpenCV for vision-related stuff. As in other topics on the TechTarget list, TensorFlow (a GitHub ML project) features prominently in its coverage (item 4). Please note: This item is labeled as "IBM Applied AI Professional Certificate (via Coursera)."

‍

Certificates vs. Certifications

‍

This is where things get interesting. Certificate programs put students through a regimented curriculum, and they usually include competency testing and project work to instill — and demonstrate — skills, knowledge and abilities.

‍

Such programs typically do NOT, however, use the psychometric approach to building, testing, and tuning exams to make sure they say something useful about the skills and knowledge of those test-takers who pass such things. Ditto for standard frameworks such as ISO/IEC/ANSI 17024 (routinely required, for example, for a cybersecurity certification to "count" toward government security job background and skills requirements).

‍

To put things a different way, while indeed the gold rush for AI/ML skills and knowledge development continues unabated, it has yet to tap into the mainstream of full-blown certification programs and offerings, especially those multi-tiered kinds of credentials with fundamentals or entry-level elements, professional or mid-level elements, and expert or master high-level elements.

‍

My best guess is that we’re still at least two years out from seeing such things appear in meaningful form. And when they do, it will take another year or two for them to roll out completely (ladders must be climbed from the bottom to get off the ground, so to speak).

‍

What about the TechTarget Top 10 List?

‍

Does that mean the aforementioned list is completely negatory? By no means. It does signify, however, that what you'll get from such things is more difficult to sell to employers, and more likely to lose its standing and value sooner, rather than later. That said, the lifetime of most technical certs — especially those like the ones for AI/ML, with a distinctly developer-oriented outlook — is usually no more than three years anyway, with many such things requiring renewals or recertification every other year.

‍

In the quickly changing AI/ML landscape that actually makes a lot of sense, so it should come as no surprise. But I would look for major players such as OpenAI, Microsoft, Google, Meta, Amazon, and others, to start moving into this certification space in the next 12-to-18 months.

‍

Pay attention to what they offer (and how those offerings work, cost, how much time they take to complete, and so forth) in formulating a long term strategy for maintaining a skills and knowledge portfolio in the areas of artificial intelligence and machine learning. There's gold in them thar hills, and the big boys should be moving the heavy equipment in soon to start extraction at scale.

‍

Stay tuned! I’ll keep you posted ...