Thursday, October 01, 2020

Artificial Intelligence Apprenticeship Program - Mission Impossible or a must-have in the modern resume ?

I was alerted by a reader who informed me of the existence of such a program. 

Unfortunately, the reader was not selected for the program but managed to tip me off to have a quick look at it. I downloaded the "field guide" which can be found here, and was quite stunned at how difficult it is to qualify for this program.

Here is my logic :

  • There is no official course to train someone to qualify for this program.
  • The self-study regime last 12 months and autodidacts must develop a range of skills from Python programming to data science to some expertise in deep learning to stand a chance to qualify.
  • The regime is hard. Even for a fresh grad who comes from some soft IS program would struggle with eigenvector decomposition to be able to start on the basics of AI work. 
  • The program requires some expertise in Microsoft Azure so experienced staff should become familiar with this as they may use a different platform in their work. I'm baffled as to why Microsoft technologies feature so prominently on this program. Even Elon Musk is not too happy at Microsoft's cornering of GPT-3. 
  • The latest update to the document was Feb 2020 so, theoretically, a person who works with the latest updated document would not be ready to qualify for the program yet. 
The entire program does not seem to make any logical sense. 

If you qualify for the program, you can achieve a training stipend of $3,500+. However, if you can successfully self-study for the course and gain skills in the document contained in the link, you should be able to ask for much more money from any start-up in existence - you are not some cheap-ass PHP/Javascript programmer that is a dime a dozen in KL. 

So it baffles me that this program is already running with actual candidates in the program.

So I have some cynical guesses as to how the program really works. Readers feel free to correct me if I am wrong.

a) One interpretation is that you really do not need to master the technology mentioned in the field guide to qualify. The selection process really selects for potential and not proficiency. Doing this will discriminate against older candidates vying for this program.

b) This is really a recruitment campaign and not an apprenticeship at all. They are looking for experienced AI personnel to be polished up in proprietary technologies like Azure to be hired by industry partners who already know which platforms to use. If so, I suspect folks with this skill set can ask for more somewhere else. ( Maybe in Silicon Valley )

If my guess is right, why should there be government support for this initiative? The benefits seem to accrue to just a few industry partners and individuals. It might even be better to reduce the stipend to $1000 and expand the program but have a higher rejection rate. 

I hope that there are actual AIAP apprentices who can shed light on this program.

Twenty years ago, the fact that it has a ridiculous rejection rate would make a 20-something version of myself rush headlong into the program without hesitation. 

This is even better than studying for the CFA.

Right now, I am too old for this shit and I would rather create my own webapp for my closed community. I think I earned the right to learn a new skill at my own pace. 

Still, the field guide is still indispensable as a "martial arts" manual for an IT professional to level up his powers, I hope the organization will launch a school like "42" in Singapore soon to build the AI ecosystem. 



  1. MS & the PAP govt is really (I mean REALLY) tight, in case you didn't know already...

  2. Oh and for non-techies, you can get revenge by investing a % of your portfolio into the established players such as MS, Amazon, Google. Or simply an AI ETF. If the propaganda is to be believed, then more big gains are still ahead.

    Btw, as a DM (the medical kind, not the D&D kind) I hope you've invested in Novo Nordisk over the years (largest pharma specialising in DM drugs). If not, just google their stock performance over the last 20 years.

    The best revenge on DM???