We often do not believe how broken things are, before we experience them ourselves. So was the case of the Labor market for (AI roles). You might be puzzled, because what more hot industry than Artificial Intelligence can you image to look for job?! Well that was my piece of mind as well, until I was subjected to 400+ job processes over the course of 6 months,
that many are bizarre to say the least. And they are completely orthogonal to your skills or weak points. They are simply HR habits and processes born before AI era and now got completely out of sync with reality.
As I didn’t want to drop these things just into well-of-disbelief for myself, I decided to release it as Wicked Hiring Advent Calendar. You can read day after day, where (AI/Tech) hiring stands now. And how not to depress yourself further, because most likely, what you go through is not your fault. And just for the record, over the 6M I describe here, I came across (turned down) and finally accepted suitable job. But the noise to signal ration is disgustingly terrible. Hiring industry is really ripe for major disruption. Read on why
(too speed up reading, sequels of this series are ordered in descending order)
PART | 6 | How it feels to REJECT GOOD JOB OFFER
Choosing right hobby, flat, life partner or new job is difficult. Not only because you need to find a good match for you, but also because you need to turn down good (yet not optimal) offer. But there is mathematical proof for optimal decision making preventing regrettably poor choice. Let me walk you through it:
Issue with us picking partners, houses or job is that set of potential options is too big to even get to know all theoretical options. Not every role is free when you are, same as not every house is on sale when you are on look-out for yours. As a result, we can assess only options that presented themselves (or we got to learn about hiddenly). But no one can meet all people of planet to choose life partner or to see all houses or employers in town. So how to pick best if you even don’t know if there are maybe better options?
Believe it or not there is algorithm designed for that bears mathematical proof of optimal choice. In expert literature this is named “Optimal Stopping criteria” but in regular population this is labeled “Secretary Choice Algorithm” (stemming from hiring for role of office secretary). You can read on this in book “Algorithms To Live By” (which I reviewed here).This optimal algorithm divides process into Scanning and Closing parts. But before you start you need to state yourself how many options do you have time (or other resources) to see at most. For example, if you unemployment benefits last 6M then you probably can only look for those 6M (of course, you can choose to invest some of your saving to go beyond whatever you are entitled to from social support or severance package.) Rather than time, you are better off counting pieces of job interviews you are willing to take, but that’s nuance.
So, if you know what your time/effort budget (let’s say you can imagine going through 50 interviews at max), then you divide this into 37% scanning and 63% closing. That means that in the first 19 cases = (37% * 50 ) you just take part, but don’t accept any of those no matter how good they seem to be. But you evaluate each option (in retrospect) and at the end of this scanning phase pick the best you have seen in these 37% scanning cases (Let’s say it was the 12th case) as benchmark.
Then you enter closing phase where you are gonna finally pick your choice. This should be the first option that you meet in closing phase that is better than best scanning option (12th case in our example). This way you are guaranteed to minimize regret and make good choice even if you at start didn’t know anything about market you are choosing from.
Now, if you have detailed knowledge of the market, you can choose to adjust the rule for you (e.g. shorter scanning). Or if there are objective criteria (e.g. Miss Universe meets you on 3rd date), you can do exception.
But the real power of this algorithm is to dare to refuse good looking offer (that is still not optimal). This happened to me as well. I was presented by very generous offer by Tim Ossenfort and his team. I had to turn it down. It was a bit awkward feeling (and there were times I felt like regretting it). I was great candidate, it took 8 stages to get to offer, and many would kill for that opportunity. Still, there were aspects that didn’t make it to Miss Universe (exception). Tim was truly outstanding leader to accept, didn’t get sour and we exchange text or two occasionally to stay in touch. And me? I found yet even better option later down the road. It was roughly 74/100 option for me, so it took lengthy and nerving months after NO to Tim and his team.
Punchline: When seeking new job, be ready to say now in early scanning phase ( ~ 37% of planned search time/cases). Once beyond scanning point take first option that is better than highest benchmark seen in scanning. This optimizes for best choice. Though you might feel down about turning down some really good (looking) choices.
PART | 5 | How I nearly fell into Labor Exploitation scheme
You likely heard of pyramid financial games or shady multi-level-marketing (MLM) schemas. Experts coined “Ponzi scheme” name to label these practices. They create (fractal-like) hierarchy and exploit individual financial contributions of lower ranks (sent to top of pyramid) who are lured with vision of them becoming top of their own sub-pyramids. Now, would you believe there are Ponzi schemes exploiting labor as well?
If you entered our Wicked Hiring Advent Calendar today, let me just briefly outline that this is 5th episode of our series on How broken the Tech Hiring market is. While in first 4 sequels we discussed apparent process issues that stem from negligence or clinging on outdated principles, today we enter more bizarre waters. As there are also nefarious and ridiculous actors on market that should not be there first place. Let me share with you experience of me coming across some in my experiment. (I entered into 432 job search processes to test how exactly Tech hiring works).
It all started with solid job ad on LinkedIn. It promoted managerial role to lead mid-size team of data, analysis and IT automation specialist. After studying the role and confirming my profile should resonate with offer, I attached my CV and hit “apply”. In about 2 days I received LinkedIn message with positive answer and invite to schedule a first call with SMS/Whatsapp in desired time. So far nothing of unusual, this is relatively common practice from headhunters.
After sending few proposed time slots, there was a silence for few hours only to receive WhatsApp from Barbara at 18:30 with request “Can we call NOW?” Well, unsocial talking hours are never green points on potential employer resume, especially if applied this abruptly and ignoring suggestions I was asked to give. As we were in middle of family dinner, I replied “I can’t talk, but will be available in about 30 min when I plan to walk the dog.”
The other side replied with “No worries, I will explain in the chat, and we can talk afterwards.” And then I received set of messages explaining that employment is:
- Fully remote
- At first I’ll work hands-on myself. The team they talked about, I will have to gradually recruit myself
- You start with 6 weeks training, where each week you will be mentored by existing team manager on skills (like video editing, website building, performance marketing set-up, etc.) and have to pass certification exam + complete tasks of given type for each week.
- No fixed salary, all variable (now that’s not even legal in most EU countries). You are directly paid for any work you do.
- This is all “work+learning+grow” and it’s combined with kind of “gamification”
This was already quite diversion both from what I wanted to do but also from what job ad was luring in. So being in chat mode, I asked: Who actually decides how much work I get? Barbara replied that first I’ll join her group and when I can recruit (and onboard) my own team members I will be getting the tasks directly from “central pile”. Who creates central pile I never learned. But I was assured that “you start with easy tasks and work yourself to more sophisticated”
The scheme was getting clear and “naked: in my eyes, so I pressed for one more information: How is one renumerated? “First 6 weeks you learn and you work on tasks proving your competence. So no pay. After 6 weeks you can get 15-20 different tasks a week (!) averaging 35-100 EURs each. So if you work hard, you can each about 1200 eur/week brutto. But your income significantly increases when you recruit your team, because you receive share of pay for tasks they complete.”
At this point, I heard all I needed. You invest 6 weeks of work without a pay, then you are pushed to do low-level tasks at mass only to recover your “invested training time” and you are expected to recruit more people for this hamster wheel for them to work for you the same way. All for about ceiling of 5K EUR pre-tax income per month, no guarantees attached.
Moreover, this “irrefusable” offer came with no official company name given, was expected to pay your “salary” over PayPal (!) and no mustrum of contract was immediately available. It was easy “No, thank you, not me”. But it struck me that there are in 21st century real labor exploitation schemes that dare to advertise (and scout) even publicly on LinkedIn.
So be aware!
PART | 4 | Most TOP MANAGEMENTS terribly MISUNDERSTAND AI
Current state of AI hiring: Existing company managers have (mostly) no AI experience, but they hope to keep their well-paid jobs by hiring AI experts under their command (to implement AI).
In first sequels of our Wicked Hiring Advent Calendar we spent time discussing the calamities of (Tech) hiring habits and processes that are visible from the outside. However, obscurity of AI hiring has also one strong intrinsic hurdle: There is complete and utter misunderstanding of How AI progress can be achieved.
As a result, many think about AI as “technology to implement”. Almost like on-premise to cloud transformation or paperwork digitalization from past. This is, of course, terribly wrong. Because those who really can build AI agents are already building them on their own and selling them as products. They don’t need (and neither dream of) joining some corporate ranks to bag for resources or fight high-politics-stakes battles. As a result, these incumbent managers end up with contracting advisors/consultants or hiring individual contributors, who know particular (party-trick) use-case that company overheard on Tech conference catering others (boasted to) had implemented. Few quarters down the road you end up with isolated islands of mini automations.
Nobody pauses to realize that if AI will be as transformational as they polish/pronounce it to be, it would probably warrant CEO or somebody on board level to be AI-deeply-literate. Or even worse, board members assume some of them will (gradually) turn into AI experts themselves. So what’s the fuss?!
The most common follow-up scenario of this misunderstanding is for CTO’s to end-up to be the “best fit” for AI challenge. Some of existing CTOs even proactively sway themselves into matter. Yet another companies seek to onboard external “CTO with AI experience”. But here reaches the AI misunderstanding its utter peak. Follow the thought with me, please: If you were leading Machine Learning or AI teams so-far, you have been Head of Data Science, Chief Data or Chief AI Officer. But not leading other SW engineers, System architects or Sys admins. If you have been practicing CTO, you have worked on Infra, CI/CD, Front-end and Back-end development, but not overseeing Computer Vision, Embeddings or NLP, and most of “your” developers never wrote single line of Python or Spark. “CTO with AI experience” is (almost) empty Venn-diagram overlap. So asking for CTO with AI track record is either forcing CDO to grow extra non-AI tech leadership arm, or CTO to sugar-coat their AI capabilities. Which one would you (rather) bet on?
But C-suites still don’t get (or chose to not get it). They throw out job offer for AI transformation leader only to test candidates on coding from Java, node.js or PHP. Only to interview about Agile or CI/CD principles, scrutinizing how to hire quickly Senior front end developers.
So let me put me straight to your face: “Dear C-level managers, if you do not have on highest level anybody who developed AI him/herself, you better get one and damn invite them to join you in your closed circle. You can’t out-hire yourself out of recent AI absence by employing AI individual contributors somewhere down in lower ranks. And if you keep fooling yourself that “AI needs to be implemented”, you will parish in dramatic shake-up after your competitors leave you 10x behind in speed of progress. By the way a bit of the hiring hint: How do you tell real AI top-manager from CTO aspiring to be one? If thought-through concept of building AI-only SW development process was not even hinted by candidate, you better pass on. You are not talking to AI-aware leader, so why waste both sides time. But first of all, stop fooling yourself AI is yet another technology to implement.
Punchline: Top managements want to run company as before, just to infuse AI in “layers below”. Real AI plan does not assume (or ask) human FTEs or vendor budgets. CTO with AI experience is misnomer.
PART | 3 | Time to answer? NO’s come 2 days earlier than Yes’es
Job offers arise continuously every day, even literally every hour. Habit of publishing the new positions in unsocial hours is really strange phenomenon worthy separate sequel of this Wicked Hiring series. (Because companies publishing new job offers in the middle of the night actually favor bots (that are ready to react any time) over humans. Thus, if late-night-owl recruiter posts new job at midnight last thing before going to bed), (s)he wakes up with only to be surprised to find 100+ profiles already reacted. Almost none of that are genuine human initiatives. So make your call.
One would foolishly assume that continuous inflow of the fresh job ads signals continuous work with the candidates as well. But anybody who has been looking for job lately can confirm for you “Those were the times …”.
We already touched in PART #2 of series upon the fact that more than half of the offers you send as candidate will be fully ghosted (no reply at all, not a single word back). But if you hoped that the other (almost) half are the good ones, well, sorry to crash your dreams.
Only about 10% of the recruiters reply within 48 hours of your application. Yes, 1 out of 10 in first 2 days. If you apply to LinkedIn job posts directly, there seems to be (automatic) SLA for negative replies for 3rd working day, because surprisingly a lot “No thank you” come exactly on 3rd day. And then, it is just long tail. Worst 10% of replies take more than 3 weeks for first contact (!). Honestly, those are probably not very promising by itself, but be my guest to give them benefit of doubt. If you exclude the automatic NO’s , median time for the first touch (proven by data) is about 5 calendar days. This is aftermath of human recruiters working in batches. Unless pushed by top management for some urgent role, it makes sense for them to review inflow of CVs in batches every x-days. For that phenomenon, hiring resembles more like seeing the doctor: Most time of your trip you spend waiting in the fore-room.
Now this is not only enough for you to bite down your nails, but also has severe implications for the rapid fire of applications you need to achieve to keep the progress somehow working for you. Because if you want to have job interview at least 2-3 times a week, compounding the ghost-rate and usual response time, you need to have continuously open 35-40 different job openings. Now stop for a while and read the previous sentence again!
Yes, job market is so broken that when you send less than 35 CV’s a week, you are unlikely to have any hiring interview scheduled for next 5 working days. They will convert in some form into talks maybe later, but that’s the Tech hiring number game now. That means to keep your pipeline hot, you need to react to 5-7 new offers per day! If you panic that “There are not 35 new positions a week opened in my location for me relevant roles” , you are probably right. But there is some remedy to that. And I share the hacks in our Wicked Hiring series in next days soon.
Punchline: Expect about 5 days for “positive” answers to come on average. Quick Nos usually arrive in 3 days. That also mean you have Nos earlier than MAYBE’s. So get mentally ready for dirt (and crashed feelings) paves the road to progress.
PART | 2 | Searching Job Is Riskier Than Lottery
One looks much calmy and with understanding on job market, if (s)he had chance to swim the river from other embankment as well. As seasoned hiring manager, I had the privilege of hiring 90+ people into Tech roles in my career so far. Therefore, I know hiring has been always numbers game. But for both sides.
For ages the candidate side of the game was running 10:3:1 rule. Out of 10 (well addressed) job applications, you would here back from 3 and you should turn at least every third CV screening call into next round progress. Well that one is definitely gone.
To make this proper experiment, I decided to store every single communication about the job opportunities I entered. As my focus was generously broad (from Global remote roles all the way to on-site job in Germany, where I stay), I used common LinkedIn platform to make the job discussions somewhat standardized and comparable.
Over the last 6M I opened the discussion with 432 roles in 373 companies. Yes, you are reading right, 400+ processes. I meticulously kept timestamps of each interaction with each of the opportunity. When did I receive first reply, what was next step, how much down the road I got? How many rounds were there for each of the role search (indicated)? Having this detailed data allows me to draw clear conclusions.
In hiring (for tech roles) we are recently in ghosting era. Why so? Well, out mentioned 400+ roles, I never received any answer (not even “After careful consideration …”) in 52% of the cases. Let me repeat it again. You apply for the job that your skills match and in more than half of the cases you don’t receive even automatic reply. Zero interaction, nothing.
If you are one of those, who’s self-confidence is not cracking stones into pieces, you most likely slide into depression. What is wrong (with me)? – ask many candidates themselves. If you belong among “challenge accepted” hardy individuals, you adjust the game. Because all of the sudden the rule changes into 65:5:1. So to get to talk to at least 1 promising hiring manager talk, be ready to send out as many as 65 applications.
Job hunting is not about chances; after all, it’s not a lottery. It’s worse than that. Because most of the national-wide lotteries have at least 4.2% (some of them as high as 11%) chances. Thus, every 1 in 24 cases wins at least something back. With Job applications you are getting first talk with hiring manager in 1 to xx cases. This is nuts. Because we still call participating in lottery hazardous behavior. So now looking for a job is even more risky than lottery participation?
Seeing this, you surely ask: “What happened? Why all the sudden? Well, having the detailed data on all cases helps me not only spotlight this problem, but also point out the actual causes of this phenomenon. We will invest into each of those root causes separate sequel of this Wicked Advent Calendar. But just to tease you a bit, they include Slavery of internal offers, Absence of dating math (this will be extra fun to read), AI cynism, Wrong level of hiring manager and Economics masquerade.
Punchline: Forget 10:3:1 rule, get mentally ready for 65:5:1 rule. Don’t try to fight it, it’s not your fault, it’s new (de)fault.
PART | 1 | WHY Job Market is BADLY broken
We often do not believe how broken things are, before we experience them ourselves. So was the case of the Labor market for (AI roles). Looking for job was (with exception of few booming years) never easy. Specially some industries and roles might have challenging to find new job opportunity. Therefore, I get friendly shoulder taps when …
… when I say I am/was looking for the AI job. “That’s so hot, there must be huge demand, and you certainly have plethora options to choose from.” Well, honestly, that’s what was my piece of mind when I had decided to take some break from previous role. But over the time, I learned that job market is really badly broken.
No matter that you hone top-notch AI skills, you have track record in the most booming industry, you can pass live coding challenges or have strong reputation or network, you will be exposed to same abyss of bizarre experience. Completely orthogonal to your skills or ambitions, as it’s all HR habits and processes that are out of sync with time and reality, to say the least. Over the course of 6 months, I came across labor scams (yes!), candidates ghosting, shields of application forms, empty hands (and promises) of head-hunters, crude misunderstanding of AI , free labor illusion and hiring practices from 90’s. All stemming from 400+ job processes I personally went through.
As I didn’t want to drop these things just into my own well-of-disbelief , I decided to release it as Wicked Hiring Advent Calendar. Over the next December days I will regularly release the daily portion of what actual labor market (for AI positions) looks like. And unbelievable stories can I promise. So stay tuned to summarizing blog or regular LinkedIn posts
Publikované dňa 30. 11. 2025.
Späť na domovskú stránku
