ACM CareerNews for Tuesday, July 22, 2025
ACM CareerNews is intended as an objective career news digest for busy IT professionals. Views expressed are not necessarily those of ACM. To send comments, please write to [email protected]
Volume 21, Issue 14, July 22, 2025
Tech Unemployment Rate Hits Lowest Yet in 2025
CIO Dive, July 7
According to the latest data from the U.S. Bureau of Labor Statistics, IT unemployment in June dropped to its lowest level yet this year, down to 2.8%. Companies across sectors added 90,000 net new tech professionals to their ranks. Almost 212,000 active job postings were added last month, with open positions available across experience levels. Tech companies, however, reduced staffing across job role types by a net 7,256 positions during the month, with the largest share of job losses in tech manufacturing. Overall, tech employment showed surprising strength for the month given recent expectations.
CIOs have adjusted tech hiring efforts this year in response to several factors, including macroeconomic uncertainty and the rise of AI. Tech unemployment has ticked up throughout much of 2025, reaching a peak in April at 3.5%. A few big-name tech giants have garnered attention for restructuring plans that have resulted in layoffs of tech workers. The wariness in tech hiring is reflective of broader caution spanning industries and roles as leaders await clearer economic signals. Still, companies are looking to fill critical gaps and tap top talent.
Click Here to View Full Article
AI-Using Managers Rely on the Tool to Decide Who Gets Promoted or Fired
HR Dive, July 9
Among the 6 in 10 managers who use artificial intelligence tools at work, 94% use them to make career-impacting decisions about their direct reports. When making personnel decisions, managers use AI to determine raises (78%), promotions (77%), layoffs (66%) and terminations (64%). More than 7 in 10 of the leaders who said they use AI to help manage their teams expressed confidence in the technology making fair and unbiased decisions about employees.
Despite the popularity of AI hiring tools, only 32% of those using AI to manage said they have received formal training on how to do so ethically, and 43% said they have received informal guidance. About a quarter said they have not received any training at all. Of the managers turning to AI, 46% said they were told to evaluate whether AI could replace the position of a direct report. Among those, 57% said they decided AI could replace the position, and 43% decided to replace the human position with AI. One big takeaway is that, while AI can support data-driven insights, it lacks context, empathy and judgment. AI outcomes reflect the data it is given, which can be flawed, biased or manipulated.
Click Here to View Full Article
Here Are the 10 Highest-Paying Jobs with the Lowest Risk of Being Replaced By AI
Entrepreneur.com, July 11
With workers increasingly facing the threat of automation, researchers recently looked at the top professions with the lowest risk of being replaced by AI. A new report showcases 10 roles that could be AI-proof. All of them meet the following criteria: high pay, high job growth, and a low risk of automation. Interestingly, all of the careers that meet the criteria are in the healthcare and applied science industries. AI can write code and crunch numbers, but it still lacks the care, judgment, and presence of humans.
Tech industry experts have been sounding the alarm about AI replacing jobs for months. To help workers avoid any impending job cuts, Resume Genius has come up with a list of10 AI-proof jobs. At the top of the list was computer and information research scientist, with a median salary of $149,910. Estimated job growth is a very robust 26%, and the so-called AI job takeover risk is just 31%. Coming in seventh on the list was operations research analyst, with a median salary of $91,290. This role has an estimated job growth of 23% and an AI job takeover risk of 42%.
Click Here to View Full Article
Outdated Cyber Hiring Practices Leading to Hiring Problems
Axios, July 15
If companies want to fix their cybersecurity hiring issues, they may need to start by rethinking their own job postings. Many cybersecurity job listings still rely on outdated titles and fail to offer the flexibility or benefits that top talent expects. This makes it harder for major companies to attract and retain skilled workers. For example, only 8% of available cybersecurity jobs at Fortune 100 companies offered remote work, and just 10% of listings mentioned mental health support.
The cybersecurity industry has long struggled with recruiting and retention. For example, the U.S. currently has enough cybersecurity professionals to fill only 74% of open roles, according to federal data. The long hours and high-stakes environments of cybersecurity jobs leave teams particularly vulnerable to burnout. As a result, organizations are increasingly acknowledging that they are going to have to evolve in order to fill all the roles available. This means finding new ways to manage teams, and getting access to a broader talent pool.
Click Here to View Full Article
This Can Help You Ace Interviews and Land a Job
CNBC, June 22
Keeping track of career wins and then putting them into a document can be one way to stand out during a job interview. This document, known as a brag doc, can record simple things like anytime you do a good job, anytime you make the life of a work colleague easier, or anytime you get kudos from management. Portions of that document can be used for emails, or as part of a slide presentation. The concept of documenting your career wins is not new, of course. But a brag doc goes one step further, by organizing aspects of your professional success into several different categories.
When putting together a brag doc, first focus on what you are good at. What roles and responsibilities come naturally to you that are harder for others? What tasks are you constantly complimented about? Once you have these outlined, you can then provide a sample of what you can deliver to employers. This section should highlight the impact of your unique skills. For example, if you are really great at cross-functional collaboration, you will likely be able to deliver efficient teams and great communication. Next, think of what you are most proud of. These are the exceptional accomplishments where you went above and beyond. For example, they might include saving the company money, or creating a process improvement that resulted in more efficiency. Think big picture in terms of your impact on your team and the business.
Click Here to View Full Article
AI and Job Cuts: Is the Picture Any Clearer?
Silicon Republic, July 15
Since becoming mainstream, generative AI technology has quickly embedded itself in every major industry, revolutionizing how we view work, efficiency and human productivity. As a result, concerns around how AI will affect the job market have become a major talking point. For now, it appears that companies investing heavily in AI are also cutting jobs. Even job search platforms are cutting jobs, amidst a broader shift into AI. The natural conclusion, then, is that AI technology leads to job cuts. But is that really the case?
AI technology, as it continues to become more advanced, offers a way for companies to become more productive, more efficient, and potentially even more profitable. However, it has yet to be proven that AI is better than human workers at performing certain tasks. For example, AI can help manage customer service workloads, but may need to be heavily supplemented by human workers in order to deliver a certain level of quality to customers. That being said, heavy interest in AI, backed by billions in investments from governments and the private sector alike, is vastly improving AI capabilities and widening its applications at an unprecedented pace. AI is often not used by itself to get tasks done, but rather in conjunction with human workers. The extra assistance from AI agents gets the job done faster, increases revenue and reduces the need for more human assistance, saving costs for companies.
Click Here to View Full Article
Strategies to Defend Against AI Bias in Your Job Search
Dice Insights, July 9
If you have applied for jobs online in the last few years, it is possible that an AI-powered recruiting tool rejected your application. Unless the AI algorithm has been specifically programmed to ignore certain indicators, the danger of hiring bias increases drastically because AI lacks crucial human judgment. New research shows the potential for significant racial and gender bias in how AI ranks resumes. In addition, a new class-action lawsuit alleges that AI systems disproportionately disadvantage older job seekers.
There are several ways to tailor your application package to avoid being rejected by an AI bot. When it comes to avoiding bias, limit any information that could trigger assumptions. Consider removing your full address and specific dates from the education and work history sections of your resume, online profiles, and digital portfolio. AI scans all of them, and together they shape the way the algorithm views you. If you have a name that is often misread or that sparks unconscious assumptions, present it in a neutral way on your resume, while still using your full name on LinkedIn to maintain consistency. You can also leverage ChatGPT, Claude, or a similar chatbot to tailor your resume and cover letter to a specific job description, or to mirror the terminology the company uses.
Click Here to View Full Article
Indeed Puts AI to Work to Help Job Seekers Find New Roles
CIO.com, July 14
Job search platform Indeed is embracing a cloud-native and data-driven digital transformation to set it up for the rapidly advancing AI era. Indeed is not just adopting AI, it is also building an agentic company where human intelligence and empathy can come together with machine intelligence. In theory, this technology transformation will give Indeed a better connection with job seekers, employers, and its own employees. The key, says top management, is having richer conversations with people, and the way to do that is with the help of AI.
Engineers at Indeed have been developing machine learning (ML) algorithms and models for many years to create the best possible matches between job seekers and employers. When they started the migration of Indeed to the cloud nearly half a decade ago, the field of generative AI was still years away from becoming mainstream. Still, they started experimenting with generative AI pilots roughly three years ago. The focus of the team on data quality made production models possible two years ago, and shortly thereafter, Indeed made the decision to deploy AI everywhere. The current plan involves training OpenAI-based models, but the company intends to remain model-agnostic and will choose whichever model to achieve the best matching of prospective employees with employers.
Click Here to View Full Article
Rethinking Distributed Computing For the AI Era
Blog@CACM, July 16
The recent emergence of DeepSeek sent shockwaves through the AI industry. DeepSeek achieved a major breakthrough in AI, and did so very efficiently. Total training costs for DeepSeek were just $5.6 million, compared to more than $100 million for OpenAI. This efficiency breakthrough reveals a fundamental mismatch between traditional distributed computing paradigms and the unique demands of AI workloads. In short, we run 21st-century AI workloads on distributed computing architectures designed for 20th-century problems. The success of DeepSeek suggests we need to rethink how we approach distributed computing for the new era of artificial intelligence.
Traditional distributed computing was designed around assumptions that no longer hold in the AI era. Consider the classic MapReduce paradigm that revolutionized big data processing. It excels at parallel problems where data can be partitioned cleanly and computations are largely independent. However, transformer architectures, the foundation of modern LLMs, exhibit fundamentally different computational patterns that challenge these assumptions. Transformer training involves dense, all-to-all communication patterns during attention computation. Every token potentially attends to every other token, creating communication requirements that grow quadratically with sequence length. This is the antithesis of the sparse, hierarchical communication patterns that traditional distributed systems handle well. As a result, the divide and conquer strategies that work so well for traditional distributed workloads can become counterproductive.
Click Here to View Full Article
I Teach Computer Science and That Is Not All
Communications of the ACM, June 13
When there is the inevitable interaction of computer science and policy in the classroom, educators should make sure to present multiple perspectives. Fairness and balance should be the norm, rather than neutrality and avoidance of non-technical topics. At the same time, courses on computing, ethics, and society should be widely available.
Computer science educators cannot avoid controversial topics, so it is important to present the issues from multiple perspectives. The focus should be the cultivation of societal responsibility. The fundamental purpose of higher education is a topic of much debate these days, sometimes framed as a choice between the purpose of truth and the purpose of social justice. Others have argued that the purpose of universities is not truth, rather it is inquiry. It could be the case that those who advocate truth, inquiry, or social justice as the purpose of higher education are getting it wrong. This is not to say that truth, inquiry, and social justice are not goals of higher education. Rather, they are means toward an end, which is the public good.
Click Here to View Full Article
Copyright 2025, ACM, Inc.