Александр Данильянц – THE HUMAN FACTOR IN AN ALGORITHMIC WORLD (страница 3)

• Psychology. Understanding oneself and others.

In my company, we have stopped looking at diplomas when hiring. Portfolios and case studies are more important to us. We care more about how a person thinks than what they have memorized. We give tasks with no right answer. We watch how the candidate searches for a solution, argues, and behaves in a dead end.

The "Centaur" Concept

In the chess world, there is a term "Centaur." It's a team consisting of a human and a computer. Research has shown that a "Centaur" beats both a pure human and a pure computer.

Why? Because the human directs the machine's computing power in the right direction, cutting off obviously dead-end branches that the machine could calculate for hours.

In business, you must become a Centaur.

Don't try to compete with AI in calculation speed. Use it as an exoskeleton for your mind.

Let AI write the draft of an email, and you infuse it with soul.

Let AI analyze the market, and you make the strategic decision.

Let AI generate 100 ideas, and you choose the one brilliant one.

The Death of the "Average"

The labor market is polarizing.

The middle is being eroded. Accountants, operations specialists, technical translators, dispatchers – these professions are transforming or disappearing.

Two poles remain:

Those who create and manage technology. (Engineers, AI Architects).

Those who work with people and meanings. (Leaders, psychologists, creators, high-level service).

If you are in the middle, you have a choice. Either dive deeper into technology (become the one who configures neural networks) or dive deeper into humanity (become the one who understands clients better than they understand themselves).

The second path is often more sustainable. Technology changes every 5 years. Human psychology changes over centuries.

Practical Exercise: Audit of Your Routine

Take a piece of paper. List all the tasks you perform during the week.

Mark those that:

• Repeat.

• Have a clear algorithm.

• Require no emotional involvement.

→ This is the "Death Zone." Try to delegate this to software or assistants.

Leave tasks where you need to:

• Persuade.

• Invent.

• Sense.

• Decide under conditions of data scarcity.

→ This is the "Life Zone." Increase the share of time spent here.

We cannot stop progress. Routine intelligence is indeed dying. But this frees us to become more creative, deeper, more alive. A machine can imitate style, but it cannot live a life. And business, ultimately, is done by people for people.

Chapter 3. ETHICS AS A COMPETITIVE ADVANTAGE

"Reputation is hard to earn, easy to lose, and almost impossible to recover." – Warren Buffett

At the start of my career in the industrial sector, I learned a lesson that cost the company millions and me several sleepless nights. We were involved in implementing a new quality control system at a production node. The algorithm, developed by external contractors, was mathematically perfect. It rejected products that didn't meet tolerances. Efficiency rose by 15%. Reporting shone green. Bonuses were paid.

But six months later, hidden problems began. Customers started returning batches. Not because the defect was obvious, but because the material behaved unpredictably under extreme conditions. The algorithm filtered out obvious defects but allowed those on the edge of tolerances, considering them "acceptable." However, human experience told technologists: "The edge is already a risk." But the system didn't account for risk; it only accounted for compliance with the figure.

When we investigated, it turned out the contractors had tuned the system to minimize the reject rate in reports to receive full payment under the contract. They optimized the system for their benefit, not for the safety of the final product. This was legally clean but ethically dirty.

This case taught me the main rule of the algorithmic world: what can be measured will be optimized. And if you haven't set ethical boundaries, the system will optimize them to zero.

Trust as a Balance Sheet Asset

In traditional accounting, there are tangible assets (machines, buildings, money) and intangible assets (patents, brands). But in the new era, the main intangible asset becomes trust.

Why? Because in a world of information overload and deepfakes, the only thing that matters is confirmed reputation.

Imagine two companies. Both sell identical financial management software.

• Company A uses algorithms to maximize profit. It collects all user data, sells it to third parties, hides subscription terms in fine print, and uses dark patterns in the interface to make cancellation difficult. Their short-term profit is 30% higher.

• Company B chooses transparency. They don't sell data. They make cancellation a one-click process. They openly discuss the limitations of their AI. Their profit is lower now.

What happens in five years?

When a data leak occurs (and it will happen to everyone), Company A loses everything. Customers leave, regulators fine them, partners turn away.

Company B receives a credit of trust. Customers say: "They warned us. They are on our side."

In the algorithmic world, the speed of spreading negativity exceeds the speed of light. One viral tweet about unethical AI behavior can destroy capitalization in hours.

Algorithmic Bias: The Hidden Enemy

We tend to think machines are objective. "Numbers don't lie." But humans input the numbers. And the data on which neural networks are trained contains all the human prejudices of the past.

In Luxoft practice, we faced the task of selecting a hiring system for a large international corporation. The client wanted to automate the initial resume screening. AI was to select the best candidates.

We audited the model on the company's historical data for the last 10 years. Who was successful? Mostly men of a certain age, graduates of certain universities.

What did the neural network do? It began automatically downgrading resumes containing the word "women's" (e.g., "women's chess club") and favoring candidates from elite universities, ignoring talented self-taught individuals.

Technically, the model worked perfectly: it predicted "success" as the company understood it in the past. Ethically, it was discrimination.

If we had launched this without human oversight, the company would have acquired an efficient tool for reproducing its own mistakes. Diversity of opinion would disappear. Innovation would stall because innovation often comes from the "other."

We insisted on incorporating an ethical filter into the contract. We forced the algorithm to ignore gender, age, and university name, focusing only on skills and test assignments. Hiring efficiency initially dropped (it was harder for the system), but hiring quality improved over the year. We found people who would have been screened out before.

Ethics as a Business Strategy, Not Charity

Many executives view ethics as an expense line. "We need to hire an ethics officer," "We need to conduct an audit," "This will slow down development."

I propose looking at it differently. Ethics is both insurance and marketing simultaneously.

Risk Reduction. An ethical business faces fewer lawsuits. Fewer fines. Fewer reputational losses.

Talent Attraction. Generation Z and Alpha do not want to work for "villains." Top specialists choose companies whose values align with their own.

Customer Loyalty. People are willing to pay more for an "honest" product. Premium market research confirms this.

Case Study: Transparency in RankBoost

In our project for promotion within neural networks, RankBoost, we faced a dilemma. Technology allows creating thousands of fake reviews, simulating activity, boosting metrics so no one notices. It's cheap and effective in the short term.

We could have offered this to clients: "We guarantee top positions in a week."