Companies aren’t just cutting jobs because of AI. Many are hiding behind it and the difference matters more than you think.

Tens of thousands of people lost their jobs last year, and the press release almost always read the same way: “As we embrace AI-driven transformation, we are streamlining our workforce to align with the future.”

It sounds inevitable. It sounds strategic. It sounds like progress.

Most of it is spin.

Of the 1.2 million job cuts U.S. companies announced in 2025, nearly twice 2024’s total, AI was cited as a reason for just 55,000, or roughly 4.5% of them, according to research firm Challenger, Gray & Christmas. Read that again. The AI narrative is dominating headlines, reshaping careers, and driving mass anxiety, yet it directly explains fewer than 1 in 20 of the actual cuts being made.

This isn’t a story about AI. This is a story about accountability, and how effectively corporations have avoided it.

Why the AI Story Is So Convenient

Let’s be clear about something first: AI is genuinely changing the nature of work. The transition is real. The long-term impact on labour markets will be significant. That’s not the argument here.

The argument is simpler and more uncomfortable: AI is being used as a narrative shield for decisions that have nothing to do with AI.

Cappelli, a Wharton professor, attributed the recent surge in layoff announcements to concerns about the state of the economy and noted a likely “bandwagon” effect, where companies see competitors cutting and start cutting too. “If it looks like everybody is cutting, then you say, ‘They must know something we don’t know,’” he said. Investors often reward cutting because “it looks like you’re doing something good. It looks like becoming more efficient.”

There it is. Layoffs aren’t just operational, they’re performative. Blaming AI adds a veneer of inevitability that pure cost-cutting doesn’t have. It tells investors: we’re not behind. We’re ahead.

And nobody wants to question it, because questioning AI feels like questioning the future.

The Four Types of Companies Hiding Behind AI

Not every company using this narrative is doing so for the same reason. Understanding the pattern helps you see what’s actually happening.

1. Companies masking poor sales performance. Starbucks’ decision to cut around 2,000 corporate jobs in two rounds was related to slowing sales and a larger turnaround effort led by its new CEO , not AI displacement. But the language around “efficiency” and “operational focus” echoes the AI restructuring playbook perfectly. Poor quarter. Restructure. Frame it as forward-looking innovation.

2. Companies chasing the AI race optics. There are organisations cutting headcount not because AI has made roles redundant but because they want to signal that they are AI-forward companies. In an investor environment where AI adoption is treated as a proxy for future competitiveness, announcing AI-driven restructuring can lift stock price regardless of whether AI is actually doing anything yet.

In many cases, AI may be serving as a “fig leaf” to cover layoffs actually driven by financial underperformance or earlier overhiring. The “fig leaf” framing is precise. It covers just enough to make the decision look principled. AI, Datacenter needs investment and that’s how they are gathering it

3. Companies hiding bad strategic bets. Block CEO Jack Dorsey publicly announced eliminating nearly 40% of his workforce reducing headcount from over 10,000 to under 6,000 explicitly crediting AI and “smaller, flatter teams.” Critics noted the move aligned suspiciously well with efforts to restore investor confidence after Block’s stock had underperformed. When business decisions need defending, AI provides the most compelling defence available in 2026. Oracle is a recent example.

4. Companies correcting strategic failures they can’t admit. Not every balance sheet disaster is acknowledged as one. Some companies made aggressive bets on markets, partnerships, or products that didn’t pay off. The AI era gives them an exit ramp that sounds like vision rather than failure.

The Transition Is Real. The Rush Is Not.

Generational technology transitions don’t happen in a quarter. They don’t happen in a year. They rarely happen in a decade without massive social and structural disruption.

The internet didn’t eliminate office jobs overnight. Automation in manufacturing took decades. A 2025 Goldman Sachs report estimated that if AI were used across the economy for all the things it could currently do, roughly 2.5% of US employment would be at risk of job loss. Not 25%. Not 50%. 2.5% and that’s the ceiling of current capability, not current deployment.

Research from Anthropic published in 2025 showed that although many work tasks are susceptible to automation, the vast majority are still performed primarily by humans rather than AI tools.

The hype cycle around AI is genuine but hype cycles create distorted decision-making. Companies feel pressure to demonstrate AI readiness before they’ve had time to evaluate it properly. The validation-before-decision framework that usually governs major organisational change has been compressed, or skipped entirely. Proof of concept has been replaced by proof of positioning.

Klarna proudly claimed its AI agents were doing the work of 700 humans in customer service, then by spring of 2025, it had backpedaled and was hiring again, having decided that in some cases human workers were preferable. This is what might be called “the Klarna Effect.”

Evaluate first. Deploy second. That’s the sequence that works. What we’re watching instead is deploy first, justify later, and let workers absorb the cost of the experiment.

What to Actually Look At

If you’re an employee, an investor, a leader, or just someone trying to understand whether a company’s layoff announcement is honest stop reading the press release. Start reading the numbers.

Look at the business performance trajectory. Has revenue been declining for multiple quarters before the AI pivot announcement? Are margins under pressure? If yes, the AI framing is almost certainly covering for financial underperformance. A company genuinely restructuring around AI capability doesn’t usually need to do it while revenue is falling.

Look at historical hiring patterns. Companies have cut 666,000 employees since 2022. Many of those same companies were on aggressive hiring sprees in 2020 and 2021, when remote work demand, pandemic-era digital adoption, and cheap capital made expansion feel unlimited. Check the hiring curve before accepting the restructuring narrative.

Look at what the company is actually building. If a company claims AI is replacing roles but isn’t investing visibly in AI infrastructure, tooling, or talent the claim is likely cosmetic. Real AI transformation creates new roles even as it eliminates others. Only 1% of services firms reported AI as the reason for laying off workers in the past six months, while 35% of services firms used AI to retrain employees and 11% actually hired more workers as a result of AI. AI adoption, when it’s real, looks more like capability expansion than pure headcount reduction.

Look at what competitors are doing. Experts noted a likely bandwagon effect companies cutting because peers are cutting, not because the operational case is there. Sector-wide layoffs that happen simultaneously aren’t usually driven by technology. They’re driven by sentiment, investor pressure, and imitation.

The Accountability Gap

The deeper issue isn’t just corporate spin , it’s that the AI narrative removes accountability entirely. When a company says “market conditions forced us to cut,” leadership owns a failure. When they say “AI is reshaping the industry,” they position themselves as visionaries navigating the inevitable.

Nobody gets held responsible. Nobody asks hard questions. The workers leave. The stock price reacts. The CEO gives a TED talk about the future of work.

The “forever layoff” pattern, layoffs that come in never-ending waves instead of a single restructuring, can “blindside” employees and damage morale in ways indistinguishable from layoffs at struggling companies. Continuous low-level cuts, each one framed as strategic AI alignment, compound into something corrosive: a workforce that has stopped trusting leadership to be honest about what’s actually happening.

That trust deficit is expensive. And no AI tool fixes it.

The Real Question to Ask

AI is changing the game. That part is true. But not every company that invokes AI in a layoff announcement is part of a genuine transformation. Some are hiding poor sales. Some are correcting hiring mistakes. Some want better numbers before an earnings call. Some just don’t want to own bad business decisions made over the last three years.

The businesses genuinely navigating the AI transition are easy to spot: they’re investing in retraining, building new capabilities alongside cuts, and communicating clearly about what’s actually driving decisions.

The others? They’re borrowing the language of progress to avoid accountability for the past.

Look at the business decisions. Look at the numbers. Look at what’s actually being built.

That’s where the real story lives.

Key Takeaways

• AI is cited in fewer than 5% of 2025 job cuts in the US, yet dominates the narrative around layoffs, a significant gap between rhetoric and reality.
• Companies are using the AI transition to mask pandemic overhiring corrections, declining revenues, bad strategic bets, and competitive optics plays.
• Real AI transformation creates new roles and retrains existing ones, pure headcount reduction with no investment in capability is a red flag, not a transformation.
• To evaluate any layoff announcement: examine the revenue trend, hiring history, and what the company is actually building, not just what it’s saying.
• The accountability gap created by the AI narrative is itself a leadership failure and ultimately more damaging to organisations than the restructuring decisions themselves.

References

1. Challenger, Gray & Christmas U.S. Layoff Report 2025. AI cited in 4.5% of job cuts.
2. Fortune “AI-washing and Forever Layoffs: Why Companies Keep Cutting Jobs”, February 2026. https://fortune.com/2026/02/10/ai-washing-and-forever-layoffs-why-companies-keep-cutting-jobs-even-amid-rising-profits/
3. CNBC “AI-washing and the Massive Layoffs Hitting the Economy”, November 2025. https://www.cnbc.com/2025/11/04/white-collar-layoffs-ai-cost-cutting-tariffs.html
4. The Conversation “Tech Companies Are Blaming Massive Layoffs on AI. What’s Really Going On?”, 2026. https://theconversation.com/tech-companies-are-blaming-massive-layoffs-on-ai-whats-really-going-on-278314
5. Goldman Sachs, AI and the Labour Market, 2025. (Cited in The Conversation above.)
6. CEIBS “Is AI Alone Driving Global Layoffs Or Is It More Complicated?”, January 2026. https://www.ceibs.edu/new-papers-columns/28376
7. Trax Tech / Yale Budget Lab “Companies Use AI as Cover for Layoffs Despite Limited Automation Evidence”, October 2025. https://www.traxtech.com/ai-in-supply-chain/companies-use-ai-as-cover-for-layoffs-despite-limited-automation-evidence
8. Fortune “9 Reasons AI Isn’t Going to Take Your Job (Yet)”, April 2026. https://fortune.com/2026/04/01/ai-layoffs-automation-productivity-finance-employment-investors-ceos/

In the world of testing, Chrome For Testing (CFT) is a significant change, transforming the way testers and developers use Chrome for their testing purposes.

Despite its significant impact, there is limited awareness in the testing community about this innovative feature introduced by the Chrome team. In this blog post, we’ll delve into the key features of Chrome For Testing, its benefits, and how it affects testers and developers using Chrome for their regular testing and automation.

The Need for Chrome For Testing:

The Challenge with the regular Chrome version is that users could only install the latest version of Chrome from the website, with no control over using specific versions or revisions. This limitation posed challenges for testers and developers in identifying and debugging issues during testing of the web application that are specific to chrome version or revision.

What is 'Chrome For Testing' :

Chrome For Testing is introduced to address this limitation, giving users control over the specific version of Chrome they want to use as per their use case. Testers now have the flexibility to choose specific Chrome versions or revisions of specific version, enhancing their testing capabilities and allowing them to identify issues with greater precision.

Key features

• Centralized Dashboard: Now users will have matching binaries for Chrome and ChromeDriver in one dashboard.
• Versioned Binaries: Chrome For Testing also provides versioned binaries for different revisions including the stable and upcoming etc, allowing users to test across various Chrome releases.
• All Channels Binaries: The availability of binaries for all channels, such as Canary, Dev, Beta, and GA, on the same Dashboard, makes testing across different Chrome channels becomes effortless and efficient.
• New Endpoints: Users will have different JSON end points that gives them the access to different revisions of old and new binaries. (Starting from chrome v113.0.5672.0 and chromeDriver v115.0.5763.0)

Bonus : It does not update automatically, just so you know.

Impact on Automation

CFT is available starting Chrome v115 and the automation community encountered a significant challenge as the Chromedriver v115 was not accessible through the conventional Chromedriver website or endpoint. Consequently, this issue resulted in their scripts failing to execute successfully.

Actions for Testers:

To overcome this obstacle and leverage the potential of ‘Chrome For Testing’, testers must take the following actions to ensure their automation remains efficient and error-free:

• Update Chrome download URLs: If you have implemented the latest download logic for Chrome, update the URLs in your scripts to use CFT binaries in your scripts (e.g. pipeline, docker etc). These Chrome browser binaries can be utilized for regular testing as well.
• Update ChromeDriver latest release end points: If you were using latest chromedriver URL: https://chromedriver.storage.googleapis.com/LATEST_RELEASE to download the compatible driver, change the endpoints as per the new JSON API endpoints provided by CFT. Detailed instructions are given here : https://github.com/GoogleChromeLabs/chrome-for-testing#json-api-endpoints.
• Use Latest Webdriver Manager: If you are using Webdriver Manager, ensure that you are using the latest version of Webdriver Manager, updated recently to handle the changes brought by CFT.
• Use Latest Selenium 4 : If you are using Selenium 4, be aware that it will cater to the Chrome For Testing change starting from version 4.11. This upgrade ensures seamless compatibility within Selenium Manager, making it the recommended course of action.

Ongoing Challenges:

• An issue with the availability of a compatible driver at this new location for default Chrome browsers with different revisions of Chrome 115.
• Fail to launch the default Chrome browser even with a matching chromedriver through automation.

I have posted these issues and workaround in my linkedIn post here

Conclusion:

While Chrome For Testing offers significant benefits for testing use cases, some teething issues and pain points may be experienced temporarily. However, in the long run, it provides substantial advantages.

These CFT binaries are introduced for testers engaged in testing and automation, offering potential for improved testing processes. It’s important for testers to be proactive in adapting to these new changes, ensuring seamless testing and automation experiences.

Let’s embrace Chrome For Testing and leverage its potential to improve our testing practices.

Note: Chrome For Testing has no impact on regular Chrome users. They can continue to download & use the default Chrome browser from the website in the same way as before.

Official links:

https://developer.chrome.com/blog/chrome-for-testing/

https://googlechromelabs.github.io/chrome-for-testing/

https://github.com/GoogleChromeLabs/chrome-for-testing#json-api-endpoints

Winter ’23 Salesforce release has been rolled out on preview orgs recently. This release brings in some important changes that may affect users in a different way.

In this blog, I am mentioning a few functional changes that I feel are important for any Salesforce user. I will also try to highlight the impact of these functional changes from a testing point of view.

1. Enhance Domain enablement

Enhanced domains are the latest version of My Domain that meets the latest browser requirements. This feature changes domain suffix to meet the latest security standards. Enhanced domains also comply with the latest browser requirements, allowing users to access Salesforce using browsers that block third-party cookies.

If you don’t postpone the enforcement using the My Domain setting, Salesforce enforces this update for sandboxes and non-production orgs in Winter ’23.

Usage and impact

If you don’t test and enable enhanced domains before the enforcement date, here are some issues that can arise:

• Users can experience errors when attempting to access Salesforce, including but not limited to Experience Cloud sites, Salesforce Sites, and Visualforce pages.
• Some embedded content stored in Salesforce no longer appears.
• Third-party applications can lose access to your data.
• Single sign-on integrations with sandboxes can fail.
• Single sign-on integrations with orgs using the *.cloudforce.com and *.database.com domain suffixes can fail.

Hence, the recommendation is to use it on your sandbox before it is enforced in Spring ‘23.

Possible impact on test strategy

The impact mentioned in the previous section is most likely gonna impact testing scenarios as well. Users may have to change the target URLs in the test and it is worth checking the navigations of different custom pages and applications on salesforce.

Also, the automation scripts need to be updated wherever these URLs are used either for navigation or for assertions.

2. Dynamic forms for Standard Object

Dynamic forms gives users more control to organize a record page. It was initially rolled out for Custom objects, but with Winter ’23 this option is extended to few Standard Objects as well Account, Opportunity and Contact.

Usage and impact

Users can configure these fields and sections wherever they want without going to the page layout editor. Also, it allows users to configure visibility rules to show and hide the fields and sections for these Standard and Custom Objects.

Possible impact on test strategy

With this change, users will have to change the strategy to test these dynamic forms pages. Considering these flexibility in configuring the field, sections and visibility rules, additional tests may be required. Automation scripts also need to be updated as per the dynamic form page configurations.

3. Einstein Search enablement

Einstein Search is an enhanced version of Global search. It refines the search better and provides users with more interactive results using personalized search results like, Natural language search and instant actionable results.

Usage and impact

This feature came a while ago, but was not available to every user but now with Winter ’23 all global search will default to Einstein search for a better user experience. However, there is an option to disable it; if users don’t want to use this feature.

Possible impact on test strategy

Since the UI has changed, it will require users to create or change existing steps/scripts and add some additional tests for the mini search layout available. Also, a few changes on the suggestion list at left should also be updated in the existing global search scripts.

4. Custom Address Fields

With Custom Address fields, users can add and retrieve address data via custom Address compound fields on standard and custom objects. Users can edit the custom address field data in records and view custom address data in list views and reports.

Custom address fields aren’t supported for Lead Conversion, Approval Process, or the address component of Flows. And Salesforce hasn’t validated custom address fields in Community Profile or Data Export.

This feature was earlier available as beta, but now generally available with Winter ‘23.

Usage and Impact

Custom Address field gives users the flexibility to use it in any object. By using compound address fields, users leverage the Salesforce functionality that optimally records address information rather than keeping them in a normal text field.

Custom Address fields Supports :

• Google Address Lookup
• State and Country/Territory Picklists
• Validation rules
• Apex Classes and Triggers
• List Views
• Reports
• Field History tracking
• Managed Packages
• Change Sets
• Apex and API
• Change Data Capture
• Custom Indexes

Possible impact on test strategy

Custom address fields behave differently than normal input fields. For example, On the View screen it may be a single field but on the edit/new screen address field it will expand to multiple fields. Also, address fields are dependent i.e. on selecting the country, the state fields are populated.

Hence, the testing approach changes to test custom address fields. And, new scripts need to be added in test cases covering these custom address fields as per the support mentioned above.

5. CPQ Salesforce Browser performance improvement

This was introduced first in Summer ’21 but has been enforced in Winter ’23. With this update, Salesforce CPQ runs faster in supported web browsers by employing Web Components V1 technology.

Usage and Impact

As a result of changes to improve the browser performance; UI changes may be observed across different browsers. However, all functionality remains intact.

The following UI changes may be observed across different browsers:

• Nested bundle products show a collapsed carrot icon while the bundle is expanded.
• [Firefox] RTE fields in quote line drawers don’t show the browser cursor.
• Row resizing change due to large text in QLE.
• Segmented line tables in LQE groups have tab size discrepancies.
• MDQ Drawer fields change to vertical spacing.
• [Safari] Quill formatting doesn’t apply when selection is collapsed (cursor mode).
• The Quote Process lookup field in the Quote Line Group shows Id instead of name.
• Custom action dropdowns on QLE Page change to vertical and horizontal spacing depending on your OS and Browser.
• The width of a Lookup field dropdown will span across the entire page, if the Lookup field is added to the Quote’s LineEditor Field Set.

Possible impact on test strategy

Generic tests may have to accommodate the UI changes mentioned above; however it is a big one from an automation testing point of view. With this setting, Shadow DOM enables in the HTML and will certainly break the tests you already have.

6. Merge Case: Master Record label changed to Principal record

Label of Master Record is changed to Principal Record on the Merge Case screen.

Winter ‘23

Summer ‘22

Usage and Impact

There is nothing specifically mentioned in the release notes around this change, it should be dealt as a UI label change.

Possible impact on test strategy

Need to update the existing tests to accommodate this UI change of label in both manual and automated tests wherever it is referenced.

7. FSL: Skills Related List changed to Service Resource Skills

About

Skills under the Service Resource object are renamed to Service Resource Skills.

Usage and Impact

There is nothing specifically mentioned in the release notes around this change, it should be dealt accordingly for the usage and reference.

Possible impact on test strategy

Users need to update the existing tests to accommodate this UI label change of Related List in both manual and automated tests used either as a reference or in the assertion value.

8. Tab Focused dialogues

It is an interesting one, where Salesforce is recommending not to enable it. This update was first made available in Winter ’20 and was scheduled to be enforced in Spring ’22, but it has been postponed to Spring ’24.

But in Winter ’23 release notes, Salesforce has mentioned “This release update has been postponed indefinitely and not enforced in its present form. Don’t enable it.”

Usage and Impact

Since there is a recommendation from Salesforce not to enable it, users can hold on to it or stop any further work in this direction.

Possible impact on test strategy

If a user is enabling it for the first time, new test scripts need to be created. However as salesforce recommended, users can hold on to it.

Introduction: As the experiences grew in the software testing domain, experts have tried to summarize and suggested following principles which are common for all testing.

1. Testing shows presence of defects: Testing can show that defects are present in a software or application but can not prove there are no defects. It means there is always a possibility of having bug in an application but testing just reduces the probability of having issues.
2. Exhaustive testing is impossible: This principle defines that testing everything is not feasible. Testing process should not focus on covering all the permutation combination of testing scenarios, but should be more focused on system requirement.
3. Early Testing: Testing process should be started early in the software development life cycle. It is helpful in finding the defects early and fixing them early that actually saves the cost of fixing the issue compare to the later stages of SDLC.
4. Defect Clustering: Testing effort should be properly observed towards the defects density. Usually few specific module contains maximum number of defects and causing most of the failures.
5. Pesticide Paradox: If the same set of tests are executed again and again, it will no longer find any defect, this is called pesticide paradox. To overcome this tests needs to be regularly reviewed and updated, keeping the focus on new area and scenario of the software.
6. Testing is context dependent: Testing is process done differently for different type of applications. for example Desktop applications are tested differently than the web applications.
7. Absence of error fallacy: Just finding the defect and fixing it will not help until the software application serves it’s requirement.

Categories

Posted on Software Testing — Manual

Originally published at http://testautomationpage.wordpress.com on April 22, 2018 by Ravindra.

What is Software Testing?

In the realm of product development, a crucial step involves quality assurance to ensure that the product aligns with its intended purpose and functions as expected.

Similarly, in the context of software products, we use the term ‘software testing.’ It refers to the process of verifying and validating the product’s quality according to specified requirements. Software testing plays a vital role in ensuring that the delivered product meets the desired quality standards.

Furthermore, the scope of software testing has expanded significantly. It now begins at the very inception of a product and continues through multiple iterations until product delivery. While the standard testing life cycle traditionally includes requirement analysis, testing planning, test case creation and execution, and closure activities, in the modern era, its scope extends far beyond these initial stages

Software Testing Origin

If we go back in time and see whether the Software Testing phase was there in the initial time of Software revolution era? The answer would be “No”.

Software engineers would typically develop a software product and deliver it to the customer or client. And somewhere in between after building the product and before delivering it to the customer, software engineer used to do a quality check by testing the product and confirming that the product is serving it’s purpose based on their understanding. But the term software testing didn’t exist as a separate entity in the beginning.

As businesses started to acknowledge the pivotal role that product quality played in customer satisfaction, revenue generation, and the overall market value of a company, management in the software industry began to take software testing seriously, And eventually it became an integral phase within the Software Testing Life Cycle.

Originally published at http://testautomationpage.wordpress.com on August 28, 2017.