[PreludeSH]

Some time ago there was code written in java, python, SQL that would make it easier without dealing directly with machine level complexities.

Even with AI, there could be some high level of pseudo instruction to the AI engine to generate code which need to learnt/mastered ? Not all people can even search effectively on google. There will always be 'special' people

[saket77]

Dear Samurai!

I am a fan of your posts and look forward to it. They are rich in information and often enlightening and full of references. Thank you for that.

Originally Posted by Samurai

Is Cognitive AI mature enough to replace IT jobs yet? Most IT work output is deterministic, and not probabilistic. No banking customer will be happy with a banking software that is accurate 99% of the time. If your salary is not deposited, would you be satisfied with the fact that 99% got their salaries? An IT customer would want their features work exactly as required, every time. You can't hide behind AI when customer yells at you.

I am not an IT person, just like to keep myself abreast of the latest on technology. On the above quoted part, even now jobs completed by humans or systems other than AI have an error component. People make mistakes and they are rectified. Many systems in form of processes or even automated ones have erred in past and they have been updated to minimize the occurrences of such errors in future. Any change in regulations require a revamp of systems. Still often errors are very much a part of results.
Then why such occurrences will be not acceptable by the same people if the same work is completed by AI? Is this a normal bias against something new coming into picture or something else? Won't AI also develop and at a rate which is much higher than brightest of the human minds so that those errors can be ironed out in future? Will ChatGPT not do much more in future than what it is doing now? And then do you think that will that affect the jobs?

Thank you in anticipation of response and for the great posts you make.

Regards,
Saket

[mayukh42]

Originally Posted by saket77

systems other than AI have an error component
...
why such occurrences will be not acceptable by the same people if the same work is completed by AI?

The error components are validated and corrected by human intervention. If we remove humans out of the equation, then bias will creep in and there will be no way to remove it (i.e. AI catching errors, AI correcting it). With every cycle, more bias will add up, and so on. Similar to Hysteresis in magnetism.

So the fundamental assumption for any learning system to evolve is that AI needs to be validated, by humans. AI should be designed to improve efficiency, not developed to replace humans.

Now, the only way out of these uncertainties is to also have AI as the customer of a AI-driven product. Then there will be no complains and everything will be perfect.

[Samurai]

Originally Posted by saket77

Still often errors are very much a part of results.
Then why such occurrences will be not acceptable by the same people if the same work is completed by AI? Is this a normal bias against something new coming into picture or something else? Won't AI also develop and at a rate which is much higher than brightest of the human minds so that those errors can be ironed out in future? Will ChatGPT not do much more in future than what it is doing now? And then do you think that will that affect the jobs?

Humans make different kinds of errors than AI.

For example, this is the kind of error AI can do.



However, you can train the AI to detect sarcasm, to some extent. But this is not about who makes more errors.

Designing, developing and deploying a software system for a given application is a systems engineering process. There is no one size fits all, it is very unlike off-the-shelf software packages. While discussing software design patterns, we have a saying that goes "There are no new problems, and there are no new solutions". They said this back in the 90s. In principle, this is true. However, the advent of newer technologies and faster hardware, has continuously allowed software engineers to solve old problems with newer and better ways. And even complex problems that were never attempted, and now being solved. In other words, software design is evolving continuously due to better tools and technology. I was designing software before internet and even local networking was available widely. A software development shop would have 20 PCs and no network at all. If you had described Kubernetes/Docker environment to me 15 years ago, I would have thought it was a science fiction. Now I design new architectures mostly to run in Kubernetes environment.

Cognitive AI is nothing but regression analysis on steroids. It needs loads and loads of data to replicate what an human can do. Once AI learns, it does it very fast. If we were doing the same thing for 30-40 years, then AI could replace us right now. Since systems engineering is a process where you are solving different problems with newer and better methods all the time, AI doesn't have enough data to work it out. The output of any systems engineering has to be deterministic, even in the first attempt. Therefore, AI is still a tool in the hands of a human designer/programmer.

[Samurai]

Originally Posted by Samurai

2) The AI may generate a million lines of code. No one can afford to review that code. Automating the code review will be an oxymoron, since AI is supposed to generate code that need not be reviewed. Regression is built into AI anyway. So, there is no need to review?
3) Therefore, code shall be verified only via testing. Who is going to write these test plans? Same AI that generated the source code, after reading the user requirements? If the source code and test plans were created by the same AI entity, why bother running the tests? AI obviously would have generated code that will pass the test plans generated by it. So, there is no need to test?

Now let's take the next logical step with this...

Computer languages were created so that human can tell computers what to do. It would be wasteful to force AI to code in human friendly language. That means ideally AI should convert user requirement directly into machine language, that only CPUs understand. Also, AI should be able to discern the intent of the user requirements exactly. This is really a hard thing. Even a human business analyst ends up asking plenty of questions to ensure that the requirements are captured correctly in written form.



That means even AI has to ask similar questions until the intent is fully understood.

AI based software development process can be like this:

1) Business analyst (BA) starts speaking to the AI, slowly with the broad strokes.
2) AI will ask numerous clarifications until the intent is clear.
3) The AI will implement and show the output right away.
4) The BA will see the output and suggest changes, which are implemented instantly.
5) The BA will describe a feature, including boundary conditions, field validation, etc., and also the tests (functional/regression/stress) that the system must pass.
6) Back to step 2.

This iterative process continues until the BA is satisfied with the full functionality. With such a process, the system can be made as deterministic as programmed by a human. That is because human is programming it, testing/reviewing it, albeit using verbal communication.

This will happen someday in future. Don't ask me how far is that day...

[Samurai]

Back to discussing salaries, Ken has an excellent article on the rise of Unhireables.

Recently Swiggy accepted a revised valuation from ₹8.2 billion to ₹5.5 billion. As a business it can't run away from reality for long. However, employees are acting more like home owners in a buyer's market. They are sticking to their peak price, that existed 18 months ago.

India’s startup workplaces confront the rise of the ‘unhireables’

India’s overfunded and overvalued startups were built on the top of overvalued talent. Today, most startups have accepted drastically lower valuations, to avoid becoming “uninvestible”. But many professionals haven’t. They risk becoming the “unhireables”

If overvalued startups are widely recognised as a problem, can overvalued employees escape the same fate?

A CEO of a prominent company estimates that, “in product and tech, 20-30% are in this category”. The Ken did a survey about this last week, and well, the responses we received from subscribers were captivating, which we’ve used to understand more about these people.

While I was not surprised by the content of the survey and the article, it also pointed out something I hadn't paid attention to... may be because I didn't hire overvalued employees myself.

the funding frenzy led to an overall decay in quality despite wages ballooning. “In the 2021 frenzy, even poor and mediocre talent got paid higher. That is not good for both the company and the individual.”

This started breaking parts of companies’ culture.

“When mediocre talent gets access to facilities like flexible hours, or remote work, or napping rooms, they abuse it. So we have to put controls on how these facilities are used,” he said.

This meant companies ended up diluting their own culture even further. “It leads to a more authoritative environment, in which only mediocre talent can survive. You alienate your high performers."

Read more at https://the-ken.com/story/indias-sta...e-unhireables/

[RodRowdyBiker]

Originally Posted by Samurai

Back to discussing salaries, Ken has an excellent article on the rise of Unhireables.

Recently Swiggy accepted a revised valuation from ₹8.2 billion to ₹5.5 billion. As a business it can't run away from reality for long. However, employees are acting more like home owners in a buyer's market. They are sticking to their peak price, that existed 18 months ago.

India’s startup workplaces confront the rise of the ‘unhireables’





While I was not surprised by the content of the survey and the article, it also pointed out something I hadn't paid attention to... may be because I didn't hire overvalued employees myself.

Read more at https://the-ken.com/story/indias-sta...e-unhireables/

What a bunch of hot garbage this article is. Just because startups as a whole went through a down round does not mean the employees also have to go through the same. The whole part about mediocre employees abusing benefits can only be said be someone who does not view employees as people but just cattle. The person who wrote the article is so out of touch with reality, and views workers as less than human. Please do not think in this way.

[Chetan_Rao]

Originally Posted by Samurai

...Unhireables...

That cuts both ways. Cash-rich start-ups overpaid for 'mediocre' talent for a few reasons:

1. They couldn't attract talent at 'reasonable' pay.
2. It wasn't their money they were burning hand-over-fist, so they didn't care if they were literally setting piles of cash on fire.
3. They didn't have the time or the inclination to do due diligence and setup proper/sustainable resource management plans.

They didn't care if their business was sustainable, only that notional 'valuation' stayed ahead of actual 'value', and hoped to find a lap for the hot potato before the latter catches up with the former and forces a correction (happening now).

It's all well and good calling high-priced employees unhireables, but they need to look at themselves too. They created this monster, as they were and are unviable employers at 'reasonable' wages for top talent, the risk:reward equation isn't in their favor compared to what established employers can offer.

Smarter employees might consider their time in the all-you-can-eat pay buffet a bonus, take a 'correction' and get back to working for stable brands. The correction may be an existential blow for start-ups without a viable resource management plan and path-to-value beyond their imaginary valuation.

P.S. How you do business is admirable, but also a rarity.

[Samurai]

Originally Posted by RodRowdyBiker

Just because startups as a whole went through a down round does not mean the employees also have to go through the same.

Why not? You make this statement, but you are not providing any justification. People had no issue doubling their pay during good days, then why refuse to come down during the bad days?

Originally Posted by RodRowdyBiker

What a bunch of hot garbage this article is... The whole part about mediocre employees abusing benefits can only be said be someone who does not view employees as people but just cattle. The person who wrote the article is so out of touch with reality, and views workers as less than human. Please do not think in this way.

Just calling the article garbage doesn't refute it. You need to explain why.

Originally Posted by Chetan_Rao

That cuts both ways. Cash-rich start-ups overpaid for 'mediocre' talent for a few reasons:
It's all well and good calling high-priced employees unhireables, but they need to look at themselves too.

These are all water under the bridge. In business one can't dwell on the past. They had much larger budget due to very low interest rates in US, allowing VCs to borrow and splurge. That is not the situation now. Now the budget is lot less and it is an employer's market again. Blaming employers for offering less is pointless.

This is just a case of not understanding how value is generated. People in sales usually know where value is generated. But others usually don't, not even finance people who know all the cost heads. In companies where Finance or Procurement has decision power to pick vendors, can cause major damage due to this ignorance. I used to deal with a large customer where procurement department did the final vendor selection. They always used to get POC/demo from us to show to their customers, and once the customer signed on, they used to switch to a cheaper vendor. That is because their incentives were based on how much money they saved in procurement. Later when the program ran into trouble due to cheap vendors, the blame was placed with the delivery team. Once they lost a $2m/year deal after a year because they switched to a vendor to save $200K one time cost. Why did this happen? Because procurement didn't understand what their customer was paying for. Customer was paying for the stability and reliability, demonstrated in the 6 month long POC. Not understanding where value is generated can lead to all kinds of incorrect assumptions.

Due to the billing hour concept in software services, it has led plenty of IT workers to assume they know their value. When an IT service company charges $50/h, it can easily be 3-4 times the salary they pay to the engineer. Therefore, a typical software engineer is very sure that the company is minting money while exploiting him/her. As if the company has no overhead at all.

[Chetan_Rao]

Originally Posted by Samurai

... Blaming employers for offering less is pointless...

Not sure if that was meant for me or a general comment but I'm not blaming anyone in isolation. Employers poached talent recklessly when they could spend like drunk sailors on shore leave, they got what they wanted at the expense of competition, often blocking talent just so the competition couldn't have them. Employees thought the reward worth the risk and took the bait, and made out like drunk sailors too. Match made in heaven, I'd say.

Now the purse strings are tighter, employers will offer what they can at reduced funding, employees with make the same risk:reward calculation and make their choices.

There's always takers for both risk and stability, beauty of a 'free' market.

[condor]

Originally Posted by Chetan_Rao

.. I'm not blaming anyone in isolation. Employers poached talent recklessly when they could spend like drunk sailors on shore leave, they got what they wanted at the expense of competition, often blocking talent just so the competition couldn't have them. Employees thought the reward worth the risk and took the bait, and made out like drunk sailors too. Match made in heaven, I'd say.

And then both suffer, the employees more than the employers. Both sides responsible. When employees started to see the higher salaries on offer, they moved. To fill these positions the back-fills had to be hired at higher salaries. Infinite loop for a while till the whole house of cards crashes down.

[Chetan_Rao]

Originally Posted by condor

And then both suffer...

Of course.

Employees become unhireable, business models become unviable. Amusingly enough, nobody seems to learn a long-term lesson. How many times in just the last decade or so have people looked at a crazy market phenomenon play out and said 'surely this won't recur?'


Have businesses learnt a lesson, or just hedging for current funding reality? I guess we'll find out.

[Turbohead]

Originally Posted by RodRowdyBiker

What a bunch of hot garbage this article is. Just because startups as a whole went through a down round does not mean the employees also have to go through the same. The whole part about mediocre employees abusing benefits can only be said be someone who does not view employees as people but just cattle. The person who wrote the article is so out of touch with reality, and views workers as less than human. Please do not think in this way.

One thing my mentor told me is that the world of IT is highly volatile where things can go up or down very quickly. Combine that with the law of demand and supply and you'll realise that when things go down for your company, you will take a hit in some form or the other. Your worth is literally based on demand and supply which is totally not under our control.

Originally Posted by Samurai

When an IT service company charges $50/h, it can easily be 3-4 times the salary they pay to the engineer. Therefore, a typical software engineer is very sure that the company is minting money while exploiting him/her. As if the company has no overhead at all.

3-4 times seems reasonable and efficient. My previous employer, a service based company, was literally billing the client 12 times my annual and still failed to be competitive in terms of profitability when compared to its peers.

[Mustang Sammy]

Revenue is flatlining as phenomena like food and instant grocery deliveries bump up against the reality of the small Indian consumer base that can actually afford these indulgences.
Cost continues to be high as economies of scale have yet to be achieved, and the frontline workers can only be squeezed so much.
That leaves tech and product folks as the ones on whom the ax is falling.

[Samurai]

Originally Posted by Chetan_Rao

Not sure if that was meant for me or a general comment but I'm not blaming anyone in isolation. Employers poached talent recklessly when they could spend like drunk sailors on shore leave, they got what they wanted at the expense of competition, often blocking talent just so the competition couldn't have them.

It is a general comment. I am saying one has to look forward, and not dwell on assigning blame. Now capital is pricy, companies have tight budget, and those employees who are still demanding 2021 salaries are stuck in the past. That is what makes them unhireable.

Originally Posted by Chetan_Rao

Have businesses learnt a lesson, or just hedging for current funding reality? I guess we'll find out.

I can guarantee it will happen again. Individuals can learn lessons, but companies can't, they are rarely allowed to think like individuals, they have quarterly results to report. I don't remember where I read this... "in a world where everyone follows the rules, the ones who break the rules will have unfair advantage". So if capital becomes cheap again, same madness will prevail again. Most will try to get that advantage.

Originally Posted by Turbohead

3-4 times seems reasonable and efficient. My previous employer, a service based company, was literally billing the client 12 times my annual and still failed to be competitive in terms of profitability when compared to its peers.

True, but if I had quoted a higher number, most would not believe it. Often, the props one has to use to acquire customer can cost more than the cost of IT salaries. If having a sales office in the right business district (say Manhattan) is key to business success, that alone can suck most of the revenue.