What Cognizant's CEO Says About Talent Strategy in the AI Era: The Truth About Graduate Hiring and Token Consumption

Cognizant's CEO Says "AI Won't Take Entry-Level Jobs" — Hiring 20,000 Graduates and the Contrarian View That "Token Consumption Is a Vanity Metric"

For Japanese companies adopting AI at their Philippine BPO sites, we draw on the remarks of Cognizant's CEO to explain, from a practical standpoint, how to redesign entry-level roles and how to build evaluation measured by "outcomes rather than usage."

---

Part 1: Why This Matters

Step 1: The Philippine Business Context (3 min)

The theme of this article is especially important for Japanese companies doing business in the Philippines. The Philippines is one of the world's leading BPO hubs (business process outsourcing — the industry that takes on work such as call centers and back-office processing), and it is a country where large numbers of young graduates are hired. A major IT firm like Cognizant, which appears in the article, also has a substantial presence in the Philippines.

The CEO of Cognizant, the central figure in the article, pushed back against the recent pessimism that "AI will take entry-level jobs" and stated that the company will again hire more than 20,000 graduates this year. This can be encouraging news for the Philippine labor market and for Japanese companies hiring locally. At the same time, his point that the substance of the work will shift toward "a role of checking AI's output and assuring quality" is an opportunity to rethink how local staff are developed.

Another point of discussion is the argument that it's a mistake to use AI usage (the amount of tokens consumed) as a yardstick for results. When advancing AI adoption at a Philippine site, the idea of evaluating people by "what results they produced" rather than "how much AI they used" is directly useful for local management.

In your Manila office, you are holding the morning meeting. The lead among your Filipino staff asks you, "The Japanese head office is telling us to cut headcount with AI — is it really going to be all right?" You answer while displaying this article on the screen: "The head of one of the world's largest IT firms is actually saying he'll increase graduate hiring. The substance of the work changes, but he won't cut people. Let's reshape our roles together too."

Step 2: Organizing the Key Points of the Source Article (5 min)

The table below was compiled independently using only the facts that appear in the source article.

Source: Fortune — "Cognizant CEO is swimming against the tide on AI: he's hiring over 20,000 graduates this year and says AI tokenmaxxing is a 'vanity metric'" (June 1, 2026)

This table was compiled from publicly available facts for learning purposes. Please refer to the original article linked above for full details.

Related: see How AI Automation Helps Philippine SMEs Solve Staff Shortages from Data Analysis to Sales.

Step 3: Comprehension Check (5 min)

Q1. What does Cognizant's CEO think about the view that AI will take entry-level jobs?

Hint: The article uses the strong phrase "overblown fear-mongering."

Q2. On what scale did Cognizant hire graduates last year?

Hint: The number "20,000" is the clue.

Q3. What did the CEO call a "vanity metric"?

Hint: It's a figure that measures how much AI was "used."

Q4. Is a technical degree always required to apply for the newly created roles?

Hint: The article cites examples of graduates in history and biology.

Q5. How does the CEO say the structure of work (the talent pyramid) will change?

Hint: Pay attention to the words "flatten" and "middle layer."

---

Related: see How AI Training Helps Philippine SMEs Build In-House AI Talent.

Part 2: Putting It Into Practice

Step 4: Implementation Steps in the Philippines (10 min)

Here are five steps for actually putting the article's theme to work at a Philippine site.

In each step, it's important to proceed together with the local managers. If the Japanese side decides everything and imposes it, local staff become more anxious. Try it first on a small task, then gradually expand the approaches that worked.

Step 5: Common Mistakes and How to Avoid Them (5 min)

Mistake 1: "Mistaking AI usage for results"

This is the mistake of rating a team as excellent because it used a lot of AI. You fall straight into the trap the article's CEO called a "vanity metric."

Bad example: Reporting to the head office, "This month our AI usage doubled from last month. That's proof productivity went up."

Good example: Saying, "Usage increased, but we'll check whether case volume and quality also improved. If results don't follow, we'll review how we're using it."

Mistake 2: "Thinking efficiency rises if you cut graduate hiring"

This is the mistake of jumping to the conclusion that you don't need graduates because you have AI. The article states that, as a role for checking output and assuring quality, entry-level talent is in fact needed.

Bad example: Stopping graduate hiring and trying to run the floor with AI alone.

Good example: Assigning graduates "the role of checking AI's output and catching errors" and developing them as the checkers.

Mistake 3: "Proceeding without explaining the role change to the local side"

This is the mistake of changing how work is done while skipping the explanation to local staff. In the Philippines, things tend to proceed on verbal agreement, which makes the later "I wasn't told" kind of miscommunication likely.

Bad example: Deciding the new division of roles on the Japanese side and finishing with just a single email.

Good example: Holding a briefing together with the local managers, conveying the meaning of the new roles to each person, and always setting aside time to take questions.

---

Part 3: Going Deeper

Step 6: Related Technical Terms (5 min)

A token (the smallest unit by which AI processes text) is the "fragment" the AI uses when it reads and writes language in fine slices. When adopting AI at a Philippine BPO site, the pricing is often determined by the amount of these tokens, so understanding it as the basic unit when budgeting is useful.

Tokenmaxxing (maximizing token consumption) is the mindset of trying to make how much AI was used look like a result. The article's CEO criticizes this, and at a Manila site too, it's important to take a stance of judging by actual results rather than being swept up in reports that boast about "the amount used."

Agentic AI (AI that thinks for itself and carries work forward) is AI that, without being instructed one step at a time, works out the procedures and moves toward a goal on its own. On the Philippine back-office floor, the use of entrusting a whole flow — from input through checking — to such AI, while people concentrate on the final check, is spreading.

Outcome-based pricing (a scheme where you receive compensation according to results) is the idea of being paid for the results produced rather than the time worked. The article's CEO says this form will become mainstream in the future, and Japanese companies contracting with Philippine outsourcers will increasingly find themselves discussing a shift from a per-hour rate to a per-outcome unit.

A workforce pyramid (the staffing composition of an organization) is the shape of an organization with many newcomers and fewer people the higher you go. The article explains that this shape flattens and the middle thins, and at Philippine sites too, designs are being considered that thicken the entry-level and senior management layers while entrusting middle-tier work to AI.

Step 7: Thinking About How to Apply This to Your Own Company (10 min)

Check whether your evaluation criteria have become "the amount used"

Something to think about: Check whether your team evaluations place weight on "the amount used," such as the number of AI uses or processing time. As the article points out, the question is whether you're measuring by results rather than volume.

Design the "checker" role to assign to Philippine graduates

Something to think about: There are situations where errors remain if you use AI's output as is. Deciding concretely who checks what, and at which stage, makes the graduate's role clear.

Explore room to shift outsourcing contracts from "time" to "outcomes"

Something to think about: If your current contract is on a per-hour rate, try writing out which tasks could be changed to a per-outcome unit. This lets your company prepare early for the future direction the article points to.

Next action: First, at your Philippine site, pick one task that uses AI and, for just one week, record "the amount used" and "the results produced" separately. If the two diverge, that's a starting point for reviewing your evaluation criteria.

---

Part 4: FAQ

Q1. In the Philippines too, will AI reduce entry-level jobs?

The article's CEO says he will, if anything, increase graduate hiring. The Philippines has a large BPO industry and abundant young talent. Rather than jobs disappearing, the realistic view is that their substance shifts toward a role of checking AI's output and assuring quality. The idea of preparing new roles, rather than stopping hiring, is useful.

Q2. If the Japanese head office tells us to "cut people with AI," how should we respond?

We recommend first proposing it as a reshaping of roles rather than a cut. In the Philippines there are procedures under DOLE (the Department of Labor and Employment), which oversees labor, and careless dismissals tend to develop into labor disputes. Explaining to the head office a path that keeps people while raising results, while citing the article's example, makes the discussion more constructive.

Q3. How should we report the impact of AI adoption to the head office?

Center your report on "what results were produced" rather than "how much AI was used." Just as the article's CEO called usage a "vanity metric," volume-only reports invite misunderstanding. It's important to select and present figures directly tied to the business, such as case volume, quality, and reduced handling time.

Q4. Is there anything to watch out for in Philippine-specific culture when adopting AI?

Be aware that things tend to proceed on verbal agreement. Always put new role divisions and evaluation criteria in writing and share them with everyone. Also, because local staff value relationships, conveying changes carefully — explaining the background at a briefing rather than handing them down one-sidedly — makes them easier to accept.

Q5. In handling the data given to AI, is there a point particular to the Philippines?

It's essential to check the rules of the NPC (National Privacy Commission), which administers the Data Privacy Act. This is especially important in BPO work that handles customer information. Decide in advance the scope of information that may be given to AI, and check with your provider's service whether you can set it so your data is not used for training.

---

Tips for Making the Most of This (3 Tips)

Replace evaluation criteria of "the amount used" with "the results produced"

The core of the article is that you must not mistake AI usage for results. Check right now whether your Philippine site is reporting AI usage as a results indicator. Simply replacing it with figures directly tied to the business, such as case volume and quality, raises the precision of your judgments.

Give graduates a clear role as the "checker"

AI's output can still contain errors, and you need people who catch them. Make a plan to develop the Philippines' abundant graduate talent as these checkers. Once the role is clear, it contributes to their own growth and stabilizes the quality of the whole organization.

Always put role changes in writing and share them with the local side

Proceeding on verbal agreement makes later miscommunication likely. Document new role divisions and evaluation criteria together with the local managers, and convey them to everyone at a briefing. Putting in the effort up front greatly reduces later trouble.

---

Bonus: How to Make Use of PH AI Works

PH AI Works is a solutions company that supports the use of AI and technology in the Philippines. On the themes of this article — "talent strategy in the AI era" and "building evaluation measured by outcomes" — you can consult us with Philippine local circumstances in mind.

As a next step, you can consult us on topics such as the following:

• AI adoption at a Philippine site and the redesign of local staff roles
• Building evaluation measured by "the results produced" rather than "the amount used"
• Handling the data given to AI and putting an operating structure in place that aligns with local rules

Please feel free to get in touch. Consultations are free.

---

References and Sources

• Fortune — "Cognizant CEO is swimming against the tide on AI: he's hiring over 20,000 graduates this year and says AI tokenmaxxing is a 'vanity metric'" (June 1, 2026)
• National Privacy Commission (NPC)
• Department of Labor and Employment (DOLE)