 |
|
|
|||||
When a new general-purpose technology emergesbe it railroads, electricity, computers, etc.companies react in predictable ways. A small minority tries to reinvent themselves around it; the majority looks first for ways to cut costs. Right now, in the middle of the most significant technological inflection since the internet, many organizations are choosing the second path. They deploy artificial intelligence to automate call centers, reduce head count in back offices, and squeeze marginal gains out of existing processes. They measure AI ROI in payroll savings and hours reclaimed. It feels rational. It feels disciplined. It feels safe. It is also the fastest way to miss the real opportunity. Innovation waves are not efficiency programs AI is not a new SaaS tool, nor is it merely a workflow enhancement. It is a rapidly evolving general-purpose technology advancing from large language models to agentic systems and toward systems that learn from interaction with environments (the so-called world models that can simulate, plan, and act). When the underlying capability is shifting every few months, optimizing for cost reduction is like trying to improve the fuel efficiency of a car while its engine is being replaced with a jet turbine. The organizations that win in moments like this do not start by asking, Where can we eliminate labor? They ask, What becomes possible that was previously impossible? Those are radically different questions. The productivity paradox should have been a warning In the early 1990s, economists puzzled over a surprising phenomenon: Computers were everywhere, yet productivity statistics stubbornly refused to reflect their impact. In a press article, Nobel laureate Robert Solow famously quipped, You can see the computer age everywhere but in the productivity statistics. That observation became known as the “productivity paradox.” At the time, many assumed the paradox was a failure of technology. My own research from that time examined why the paradox appeared at all, showing that productivity measurement lags widely behind actual transformational change and that the mechanisms of value creation were not captured by conventional metrics. The explanation was obvious only in hindsight. The gains were diffuse, uneven, and entangled with organizational change. Companies had digitized old processes instead of redesigning them. Today we are watching the same pattern unfold with AI. AIs impact wont show up neatly in cost metrics Artificial intelligence does not produce clean, linear productivity gains that fit neatly into quarterly dashboards. Its effects are asymmetrical. One employee using AI effectively may outperform 10 peers. Another may misuse it, degrade quality, or even endanger our corporate cybersecurity plans. Some teams redesign workflows entirely, while others bolt AI onto legacy processes and call it transformation. The result is what researchers now call measurement myopia: the inability of traditional metrics to capture improvements that are real but not directly tied to hours worked or cost saved. Trying to measure AIs value solely through immediate cost savings is like trying to measure the value of electricity by counting candles not purchased. Efficiency is the comfort strategy, but not the opportunity one Cost-cutting is attractive because it fits existing governance structures. CFOs understand it. Boards reward it. Metrics are clear. Exploration is messier. It requires experimentation without guaranteed returns. It demands a tolerance for failure. It produces intangible benefits before visible ones. But in periods of fast innovation, efficiency is often the comfort strategy of laggards who dont yet understand what is happening. If AI is treated primarily as a head-count-reduction tool, organizations will optimize the present and sacrifice the future. They will standardize mediocrity instead of discovering leverage. Exploration, not exploitation, builds capability Advocating exploration does not mean abandoning discipline. It means redefining it. Leaders should be asking: What new products can we build with AI-native capabilities? What decisions can we delegate to systems that learn from feedback? How can we redesign workflows, not just automate them? Companies should mandate controlled experimentation across teams, not restrict AI usage to narrow cost-justification pilots. They should treat AI like an R&D posture rather than a shrink-the-budget posture. Organizations that treat AI as an exploratory layerencouraging teams to test, prototype, recombine, and rethink workflowswill build institutional fluency. They will develop internal champions. They will uncover unexpected value that no top-down cost initiative would have surfaced. The real risk isnt overspending. Its under-imagining The greatest risk in this moment is not overspending on AI. It is under-imagining it. Companies that chase short-term efficiency gains may report modest improvements and declare success. Meanwhile, more ambitious competitors will redesign their operations, products, and customer experiences around capabilities that didnt exist two years ago. Over time, the gap will not be a few percentage points of margin. It will be strategic. In periods of rapid technological change, survival does not belong to the most efficient. It belongs to the most adaptive.
Category:
E-Commerce
For decades, formative assessment has been a silent engine for learningpowering insights about student progress and worker readiness. But lets be honest, in a world where technology is evolving faster than human skills, its time to ask questions about traditional teaching and learning models, and in many cases, modernize them. So, lets talk about formative assessment in the age of AI. Formative assessment is the ongoing process educators and workplace trainers use to understand where students are in their learning and how to adjust instruction accordingly, through homework, essays, quizzes, and short writing assignments. Eighty percent of educators rate formative assessment as extremely or very important. Unfortunately, but understandably, the arrival of generative AI has made it difficult for instructors to determine what students genuinely understand, as AI tools can produce polished work instantly. THE FUTURE OF ASSESSMENT DESIGN While administrative policy can help address improper AI use, the real potential for progress comes from evolving assessment design itself. When assessments are built to prioritize the thought process rather than just the product, AI becomes far less disruptive and far more beneficial. Asking students to make their thinking visiblethrough reflections, revisions, or short explanations of how they approached a taskrestores the instructional signal that AI might otherwise obscure. For educators, this means being able to spot misconceptions earlier, tailor feedback more precisely, and differentiate support without increasing workload. This shift isnt about adding complexity. If anything, its about adding clarity. And its an opportunity to modernize assessment in ways that mirror the world students are entering. In most professional environments, AI assistance is not only allowed; it is expected. Success comes from knowing how to use these tools responsibly: checking sources, critiquing the quality of generated outputs, and adapting insights to novel contexts. Assessments that emphasize reasoning, analysis, and the ability to apply knowledge to new situations better reflect these real-world demands. They prepare students not just to complete tasks, but to think with AI in ways that enhance their learning and judgment. TEACHER BENEFITS For instructors, thoughtfully integrating GenAI within formative assessment can also reduce friction. Welldesigned tools can automate repetitive tasks such as generating varied practice items, suggesting targeted feedback language, or providing examples at different proficiency levels. This allows educators to spend more time on the highvalue interactions that deepen learning and provide individualized support. In an era of rising expectations and constrained capacity, that shift matters. There is another often overlooked benefit: insight. When AI helps surface patterns in student work, it gives educators a clearer starting point for instruction. With better visibility, teaching becomes more adaptive, and learning becomes more personalized. This is especially powerful in large classes, hybrid formats, or virtual learning environments where realtime insight can be harder to access. Recent Pearson research reveals strategies for schoolteachers and higher education instructors to evolve their formative assessments in a GenAI era. Of course, none of this happens automatically. Bold, collaborative action is required across school and highereducation leadership, administrators, and policymakers to ensure formative assessment evolves in meaningful and sustainable ways. Together, these groups play a critical role in providing a clear AI strategy, supporting educator training, and shaping an ecosystem that aligns curriculum, instruction, and assessment with responsible GenAI use. This transition also requires assessments that reward thoughtfulness over polish, reasoning over rote, and application over replication. And it requires a shared understanding that AI is not a shortcut to learning but a catalyst for insightone that can elevate the quality of teaching when used intentionally. A LOOK AHEAD The future of formative assessment isnt about outsmarting AI or pretending it doesnt exist. Formative assessment must remain fundamental to good teaching and effective learning. Ensuring AI strengthens reflection, feedback, and understanding will allow it to become a partner, rather than a substitute for learning. With thoughtful action, the integration of AI into teaching and learning can move us closer to what education has always aspired to deliver: deeper learning, clearer understanding, and better outcomes for every learner. Tom ap Simon is the president of higher education and virtual learning at Pearson.
Category:
E-Commerce
IBM stock was down 10% on Monday afternoon after Anthropic published a blog post about how its Claude Code tool can be used to modernize software written in the COBOL language, which handles large-scale batch transactions. Many of the software systems used by the federal government, banks, and airlines are written in COBOL (“Common Business-Oriented Language”), and most of those systems run on IBM mainframes. IBM also generates revenue from servicing, modernizing, and consulting on those mainframes. If COBOL code were converted to a more modern language, the systems would likely migrate to newer cloud servers. But modernizing COBOLwhich was developed 67 years agois a slow and expensive process, largely because the code can be difficult to understand and easy to break. It often reflects decades of institutional knowledge and workflows, and is frequently poorly documentedmeaning its true intent can only be uncovered through close analysis. These challenges are compounded by the shrinking pool of programmers who know COBOL. Most university computer science programs no longer teach it. Anthropic says this analysis phase is the most time-consuming and costly. Thats where Claude Code comes in. The tool can uncover and document workflows hidden within the code, identify dependencies across different parts of a code base, and give engineers insights into how to redesign systems. With AI, teams can modernize their COBOL code base in quarters instead of years, the company writes in the blog post. IBM says the analysis phase is not the hardest part. “Translating COBOL is the easy partthe real work is data architecture redesign, runtime replacement, transaction processing integrity, and hardware-accelerated performance built over decades of tight software and hardware coupling,” an IBM spokesperson said in an email. “That is the problem IBM has spent decades learning to solve, and AI is the most powerful tool we have ever had to do it.” COBOL was developed in 1959 via a public-private partnership that included the Pentagon and IBM, with the goal of creating a universal, English-like programming language for business applications. But private-sector companies have largely moved away from it. The code is difficult and costly to maintain and was designed for batch processing, making it poorly suited for modern cloud-based and real-time applications. (Anthropic and IBM did not immediately respond to requests for comment.) The U.S. government, despite repeated modernization efforts, continues to rely on COBOL-based mainframe systems to manage a wide range of financial transactions, including tax payments and refunds, Social Security benefits, and Medicare reimbursements. Anthropics blog post comes in the middle of a separate dispute between the company and the government. Anthropic CEO Dario Amodei is expected to meet with Defense Secretary Pete Hegseth to explain why the company has not removed all safety guardrails from its AI models for Pentagon use. Anthropic has drawn the line at providing AI for autonomous weapons or systems that mass-surveil American citizens. At the moment, Anthropics models are the only ones approved for government use with classified information. Anthropic says its blog post about COBOL modernization is unrelated to its friction with the government. “The timing here isn’t related to a new product or any events,” a company spokesperson said in an email. “This is part of an ongoing series of content we’ve been publishing around code modernization and Claude Code.” And Anthropic’s blog post may not be the only factor affecting IBMs stock. Investor concerns about the speed and breadth of AI deployment have depressed enterprise software stocks more broadly. The market may also be reacting to uncertainty surrounding new global tariff announcements, which could affect tech companies and their supply chains.
Category:
E-Commerce
All news |
||||||||||||||||||
|
||||||||||||||||||