Saturday, June 06, 2020

It's going to be hard to retool middle-aged PMETs.

To level up in machine learning, I had been pushing myself to pick up the mathematical foundations to understand this class of algorithms better. For the hardcore AI audience that know a lot more than I do, I am currently teaching myself Eigen and Singular Value Decomposition. The aim is to figure out what an "eigen-company" is, some kind of alien AI-generated financial statement that can be combined with others in a linear manner to generate all 30 STI companies that form the training data.

Needless to say, this is almost science fiction to a new learner like me. It is largely pointless to execute a few lines of code but not be able to understand the true meaning of these algorithms. So I've been attending Youtube lectures on matrix mathematics where MIT lecturers are able to break it down for laymen like me to get a grasp of how things are really done. I am really glad I did Electrical Engineering 20 years ago. It is like the liberal arts of engineering degrees where we are not taught anything specific but a foundation in maths and physics to solve problems in the far future.

Combining my new insights into AI and financial markets, I am now quite worried about government attempts in trying to retool middle-aged workers like me. Why I think the government will need to press on with this to reform the economy, I strongly believe due to my own personal struggles that a lot of Gen-X workers may not be saved.

a) The guys who can retool for the future economy may not need help, the ones that do might not benefit from training in the first place. 

If you think about it, the Tier-S scholars do not need help because they have iron-rice bowl jobs and a high CEP guaranteeing them a great stable life. The Tier-A guys in my cohort, MNC elites who are slightly below the scholars, have settled into directors jobs in MNCs so are busy doing the retrenchment of other people.

Problems begin with Tier-B or middle-tier Gen-X guys like me who have some job insecurity and may not even have the mathematical foundation to seriously understand the impact AI has on the economy. Yup, we PMETs who cannot find a niche in the companies that we work for.

For me, I was lucky. I found a second wind as an investment trainer, but my rice-bowl is only as secure as my last batch of training I can conduct, hence the AI preparation I am taking on my own.

b) No training is adequate for future-proofing the individual.

I think I am pretty good at training having survived 13 batches, and yet training myself is the biggest challenge I ever faced. A lot of my Python training was based on my previous life as an IT guy. Moving into data science was not too hard because I completed an entire suite of data science courses in R prior to law school.

My current regime is hard. I spent days cleaning financial data on my own often studying the shape of the data I get from different data sources. This requires getting into JSON which I was unfamiliar with. Then I have to understand the machine learning to do really simple stuff like NLP and clustering algorithms on it.

I cannot imagine how a program that covers the tech programming and domain speciality can be done within a short time to retool someone without the person starving to death before the end of the course.  If he does succeed in graduating from the program, I expect the half-life of information to be very short - how-to earn sufficient ROI from training investments?

Also IT trainers, being introverts, hardly make great trainers. I know, because of the sheer amount of maths lectures I been through the past week.

c) Even if you are fully trained, who will hire you?

Who is willing to hire an army of freshly trained AI-proficient 40-something-year-olds, even if that is even possible?

My solution to that problem is that I am self-employed. I can meticulously use AI to attack financial datasets and monetise the results in my training programmes. Our fund managers are too busy intellectually masturbating themselves into thinking they are the next Warren Buffett. Who is interested in the decomposition of company statements into their eigen counterparts?

Who knows or gives a shit about the skills I developed other than myself?

I can't imagine what kind of company will want this kind of expertise to provide a permanent job. Even if a smart Tier-2 Gen-X guy can master the training, choosing the wrong domain can extend the timing of the job search.  In that sense, I was stubborn enough to refuse to use datasets provided by textbooks ( they all involve orchids and plants). But making the code work on financial data was hell.

I doubt the government can seriously think that they can retool my generation for the sexiest skills like IoT, Fintech or AI. More likely the government will stick to the low hanging fruit and find short training stints to allow my generation to take on temporary job roles to put food on the table. This means extended periods of underemployment and continuous retrenchment.

As I cannot foresee the company of the future, I have no idea what will happen to Millenials when they hit their mid-40s.

Without investment income, maybe they can remain single forever, live in 2-room flats and always hustling for short term contracts until they can get their CPF money out.

That's assuming they have CPF money in the first place.










2 comments:

  1. It's not to create 40 year olds who are proficient in machine learning. It's to pretend that the government is trying to fix the employment problem while letting course providers earn money.

    ReplyDelete
  2. Hi Chris,

    It's hard. However, it does not mean that it is not possible. Put in the best effort and don't bother about the things beyond of the circle of influence.

    WTK

    ReplyDelete