The Shifting Career Ladder — from nafez.substack.com by Nafez Dakkak AI is changing how work works and quietly removing the pathways through which young people learn to become experts.
AI is reshaping how people build skills, enter professions, and move along the career ladder and through the labour market.
In this conversation, I sit down with Matt Sigelmen founder of LightCast and now the President of Burning Glass Institute. Matt has dedicated his career to understanding the labor market and helping society improve the connections within in it.
Matt and I explore why people and opportunities are often only “a few skills apart,” why entry-level work may be losing its traditional role as the first rung of expertise, and why schools, universities, and employers now need to rethink the pathways that turn potential into mastery.
Educators need to be deeply aligned with what these changes are, and they need to shift the AI discourse from “how” questions to “what” questions. What do we need to teach? What do we need to keep in the curriculum?
What if the salvation of our democratic enthusiasm doesn’t come from a political revolution, but a technological one? Enter the thought experiment of Dollar Legal.
Imagine a ubiquitous, consumer-facing legal tech platform—accessible from any smartphone—that commoditizes basic legal defense and assertion of rights. It is AI-driven legal agency for the masses, costing no more than a cup of coffee, or perhaps literally a single dollar.
Dollar Legal wouldn’t exist to litigate complex corporate mergers; it would exist to handle the agonizing friction of everyday survival. It instantly analyzes a notice to quit, drafts a legally sound response asserting warranty of habitability defenses, and files it electronically. It forces the bureaucratic machine to pause. It translates the raw, terrifying human experience of a legal threat into the cold rules and procedures that the justice system understands.
And the price, $1.00, at sufficient scale, would support the operation and maintenance of the service.
By deploying Dollar Legal, we fundamentally shift the balance of power. The democratization of legal leverage changes the calculus for bad actors who rely on the silent default judgments of the unrepresented.
From DSC: I used to be able to bring up Firefly on the web and use it “free” of charge — I didn’t have to go purchase tokens or credits. (I was actually paying for the Adobe Creative Cloud Pro suite of tools…so it wasn’t really free.)
But the other day I was trying to figure out what the latest pricing is at Adobe with that suite of tools and the use of credits for AI-based features. They say Adobe Creative Cloud Pro users get 4000 credits a month. Well, I have that suite and I’m still getting prompted to purchase credits. Firefly for individuals runs from $9.99 (2,000 credits/month) to $139.91 per month (50,000 credits per month). Not inexpensive, right? Below are other items along these lines.
The Era of Affordable AI Is Over. What Comes Next? — from builtin.com by Ameya Kanitkar AI providers are shifting to usage-based billing for their services. AI fluency is more important now than ever to make the most of your tools to avoid unnecessary spending.
Summary: The era of cheap, flat-rate AI is ending as providers shift to usage-based billing. Every prompt now carries a direct cost, turning casual use into major budget risks, as seen when Uber depleted its 2026 AI budget in four months. Leaders must now track real-time value and token efficiency.
For a brief window, companies had access to the most transformative technology in a generation at the cost of a streaming subscription. Tools like ChatGPT put AI within reach of anyone with a browser and time for experimentation, while GitHub Copilot came in at just $10 a month, with token costs remaining relatively low. In the beginning, experimentation felt cost-effective, easy and relatively low-risk.
But that era is ending, and the bill is coming due faster than a lot of enterprise leaders anticipated.
Tokenomics will be a hot topic of discussion on university campuses because, as Marc Watkins notes in this article, there is no realistic path forward to providing all students with access to advanced AI.
And now there is a third layer emerging. Institutions are waking up to a systems-level question they are likely not remotely prepared for. Who pays for AI? How are budgets managed when there are unclear token consumption pricing models? How is AI procured? Who decides what tools get used and by whom and who gets access and at what level?
Online learning platform Coursera is taking a page straight out of TikTok’s playbook. The company has launched a new AI-powered feed designed to serve short-form educational content in a scrollable, personalized format, signaling a major shift in how digital learning platforms may try to keep users engaged.
The feature introduces bite-sized video lessons, clips, and explainers curated through artificial intelligence based on a user’s interests, learning habits, career goals, and previous course activity. Instead of committing to hour-long lectures or full certification programs upfront, users can now discover short educational snippets designed to make learning feel more casual, accessible, and addictive.
Users scroll through a feed of short educational videos and AI-curated learning moments covering topics ranging from coding and business to AI, productivity, data science, and personal development.
But the first wave of rigorous research on those policies — including two major U.S. studies — does not point neatly in one direction. Some studies have found modest academic gains from cellphone restrictions. Others have found little to no effect on test scores, even when student phone use dropped sharply. Some studies suggest benefits for low-achieving students, others for girls, and still others for boys. In some places, attendance or student well-being improved. In others, they didn’t.
The scientific process can be messy. Cultural differences may explain why the bans are more effective in some places than others. But almost any education reform will get different results in different places, even within a single country. And the current confusion may also stem from how difficult it is to study cellphone bans in the real world.
Ideally, researchers would randomly assign some students to surrender their phones while others kept them, and then measure the effect on academic performance — the equivalent of a clinical trial for an education policy. But those experiments are difficult to enforce in schools, and so far only one study, conducted among college students in India, has attempted a randomized controlled trial. It produced a notably strong improvement in course grades for lower achieving students.
Instead, most studies rely on rougher real world comparisons that capture only partial effects of cellphone restrictions.
The quest to build a better AI tutor— from hechingerreport.org by Jill Barshay Researchers make progress with an older ed tech idea: personalized practice
One promising idea has less to do with how an AI tutor explains concepts and more with what it asks students to practice next.
A team at the University of Pennsylvania, which included some AI skeptics, recently tested this approach in a study of close to 800 Taiwanese high school students learning Python programming. All the students used the same AI tutor, which was designed not to give away answers.
But there was one key difference. Half the students were randomly assigned to a fixed sequence of practice problems, progressing from easy to hard. The other half received a personalized sequence with the AI tutor continuously adjusting the difficulty of each problem based on how the student was performing and interacting with the chatbot.
The idea is based on what educators call the “zone of proximal development.” When problems are too easy, students get bored. When they’re too hard, students get frustrated. The goal is to keep students in a sweet spot: challenged, but not overwhelmed.
The researchers found that students in the personalized group did better on a final exam than students in the fixed problem group. The difference was characterized as the equivalent of 6 to 9 months of additional schooling, an eye-catching claim for an after-school online course that lasted only five months. … To address this, Chung’s team combined a large language model with a separate machine-learning algorithm that analyzes how students interact with the online course platform — how they answer the practice questions, how many times they revise or edit their coding, and the quality of their conversations with the chatbot — and uses that information to decide which problem to serve up next.
From DSC: The types of postings/articles (such as the one below) make me ask, are we not shooting ourselves in the foot with AI and recent college graduates? If the bottom rungs continue to disappear, internships and apprenticeships can only go so far. There aren’t enough of them — especially valuable ones. So as this article points out, there will be threats to the long-term health of our talent pipelines unless we can take steps to thwart those impacts — and to do so fairly soon.
To me…vocational training and jobs are looking better all the time — i.e., plumbers, carpenters, electricians, mechanics, and more.
Can New Graduates Compete With AI? — from builtin.combyRichard Johnson The increasing adoption of AI automation is compressing early-career jobs. How should new graduates get a foothold in the economy now?
Summary: AI is hollowing out entry-level roles by automating routine tasks, eliminating a rung on the career ladder. New graduates face intense competition and a rising skill floor. While firms gain short-term productivity, they risk a long-term talent shortage by eliminating junior training grounds.
Conversations about AI have covered all grounds: hype, fear and slop. But while some roll their eyes at yet another automation headline, soon?to?be graduates are watching the labor market with a very different level of urgency. They’re entering a world where the old paradox of needing experience to get experience is colliding with a new reality: AI is absorbing the standardized, routine tasks that once defined entry?level work. The result isn’t just a shift in job descriptions or skill-requirements, but rather a structural reshaping of the career pipeline.
Entry-level workers face an outsized disruption to their long-term career trajectories. They have the least buffer to adapt given their lack of relevant job market experience and heightened financial pressure to secure a job quickly with the student-debt repayment periods for recent graduates looming.
Momentum early in one’s career matters, and the first job on a resume shapes future compensation bands and opportunities. It also serves as a signal for perceived specialization or, at minimum, interest. Losing that foothold has compounding effects to one’s career ladder.
Anthropic has launched the Anthropic Institute, a new research effort focused on the biggest societal challenges posed by more powerful AI systems.
The institute will study how advanced AI could affect the economy, the legal system, public safety, and broader social outcomes.
Anthropic co-founder Jack Clark will lead the institute in a new role as the company’s head of public benefit.
The new unit brings together Anthropic’s existing red-teaming, societal impacts, and economic research work, while adding new hires and new research areas.
Department of Labor releases AI Literacy Framework: The framework defines AI literacy as competencies for using and evaluating AI responsibly, with a primary focus on generative AI in the workplace.
Framework outlines five core AI literacy areas: These include understanding AI principles, exploring real-world uses, directing AI effectively, evaluating AI outputs, and using AI responsibly.
Guidance for workforce and education systems: The framework also provides training principles and recommendations for workers, employers, education providers, and government agencies to expand AI education and training.
Jim VandeHei’s note to his kids: Blunt AI talk — from axios.com by CEO Jim VandeHei Axios CEO Jim VandeHei wrote this note to his wife, Autumn, and their three kids. She suggested sharing it more broadly since so many families are wrestling with how to think and talk about AI. So here it is …
Dear Family: I want to put to words what I’m hearing, seeing, thinking and writing about AI.
Simply put, I’m now certain it will upend your work and life in ways more profound than the internet or possibly electricity. This will hit in months, not years.
The changes will be fast, wide, radical, disorienting and scary. No one will avoid its reach.
I’m not trying to frighten you. And I know your opinions range from wonderment to worry. That’s natural and OK. Our species isn’t wired for change of this speed or scale.
My conversations with the CEOs and builders of these LLMs, as well as my own deep experimentation with AI, have shaken and stirred me in ways I never imagined.
All of you must figure out how to master AI for any specific job or internship you hold or take. You’d be jeopardizing your future careers by not figuring out how to use AI to amplify and improve your work. You’d be wise to replace social media scrolling with LLM testing.
What if the biggest change in education isn’t a new app… but the end of the university monopoly on credibility?
Jensen Huang has framed AI as a platform shift—an industrial revolution that turns intelligence into infrastructure. And when intelligence becomes cheap, personal, and always available, education stops being a place you go… and becomes a system that follows you. The question isn’t whether universities will disappear. The question is whether the old model—high cost, slow updates, one-size-fits-all—can survive a world where every student can have a private tutor, a lab partner, and a curriculum designer on demand.
This video explores what AI has in store for education—and why traditional universities may need to reinvent themselves fast.
In this video you’ll discover:
How AI tutors could deliver personalized learning at scale
Why credentials may shift from “degrees” to proof-of-skill portfolios
What happens when the “middle” of studying becomes automated
How universities could evolve: research hubs, networks, and high-trust credentialing
The risks: cheating, dependency, bias, and widening inequality
The 3 skills that become priceless when information is everywhere: judgment, curiosity, and responsibility
From DSC:
There appears to be another, similar video, but with a different date and length of the video. So I’m including this other recording as well here:
What if universities don’t “disappear”… but lose their monopoly on learning, credentials, and opportunity?
AI is turning education into something radically different: personal, instant, adaptive, and always available. When every student can have a 24/7 tutor, a writing coach, a coding partner, and a study plan designed specifically for them, the old model—one professor, one curriculum, one pace for everyone—starts to look outdated. And the biggest disruption isn’t the classroom. It’s the credential. Because in an AI world, proof of skill can become more valuable than a piece of paper.
This video explores the end of universities as we know them: what AI is bringing, what will break, what will survive, and what replaces the traditional path.
In this video you’ll discover:
Why AI tutoring could outperform one-size-fits-all lectures
How “degrees” may shift into skill proof: portfolios, projects, and verified competency
What happens when the “middle” of studying becomes automated
How universities may evolve: research hubs, networks, high-trust credentialing
The dark side: cheating, dependency, inequality, and biased evaluation
The new advantage: judgment, creativity, and responsibility in a world of instant answers
How Your Learners *Actually* Learn with AI— from drphilippahardman.substack.com by Dr. Philippa Hardman What 37.5 million AI chats show us about how learners use AI at the end of 2025 — and what this means for how we design & deliver learning experiences in 2026
Last week, Microsoft released a similar analysis of a whopping 37.5 millionCopilot conversations. These conversation took place on the platform from January to September 2025, providing us with a window into if and how AI use in general — and AI use among learners specifically – has evolved in 2025.
Microsoft’s mass behavioural data gives us a detailed, global glimpse into what learners are actually doing across devices, times of day and contexts. The picture that emerges is pretty clear and largely consistent with what OpenAI’s told us back in the summer:
AI isn’t functioning primarily as an “answers machine”: the majority of us use AI as a tool to personalise and differentiate generic learning experiences and – ultimately – to augment human learning.
Let’s dive in!
Learners don’t “decide” to use AI anymore. They assume it’s there, like search, like spellcheck, like calculators. The question has shifted from “should I use this?” to “how do I use this effectively?”
So where do you start? There are many agentic tools and platforms for AI tasks on the market, and the most effective approach is to focus on practical, high-impact workflows. So here, I’ll look at some of the most compelling use cases, as well as provide an overview of the tools that can help you quickly deliver tangible wins.
…
Some of the strongest opportunities in HR include:
Workforce management, administering job satisfaction surveys, monitoring and tracking performance targets, scheduling interventions, and managing staff benefits, medical leave, and holiday entitlement.
Recruitment screening, automatically generating and posting job descriptions, filtering candidates, ranking applicants against defined criteria, identifying the strongest matches, and scheduling interviews.
Employee onboarding, issuing new hires with contracts and paperwork, guiding them to onboarding and training resources, tracking compliance and completion rates, answering routine enquiries, and escalating complex cases to human HR specialists.
Training and development, identifying skills gaps, providing self-service access to upskilling and reskilling opportunities, creating personalized learning pathways aligned with roles and career goals, and tracking progress toward completion.
Here’s what’s shaped the AI-education landscape in the last month:
The AI Speed Trap is [still] here: AI adoption in L&D is basically won (87%)—but it’s being used to ship faster, not learn better (84% prioritising speed), scaling “more of the same” at pace.
AI tutors risk a “pedagogy of passivity”: emerging evidence suggests tutoring bots can reduce cognitive friction and pull learners down the ICAP spectrum—away from interactive/constructive learning toward efficient consumption.
Singapore + India are building what the West lacks: they’re treating AI as national learning infrastructure—for resilience (Singapore) and access + language inclusion (India)—while Western systems remain fragmented and reactive.
Agentic AI is the next pivot: early signs show a shift from AI as a content engine to AI as a learning partner—with UConn using agents to remove barriers so learners can participate more fully in shared learning.
Moodle’s AI stance sends two big signals: the traditional learning ecosystem in fragmenting, and the concept of “user sovereignty” over by AI is emerging.
For Cogniti to be taken seriously, it needs to be woven into the structure of your unit and its delivery, both in class and on Canvas, rather than left on the side. This article shares practical strategies for implementing Cogniti in your teaching so that students:
understand the context and purpose of the agent,
know how to interact with it effectively,
perceive its value as a learning tool over any other available AI chatbots, and
engage in reflection and feedback.
In this post, we discuss how to introduce and integrate Cogniti agents into the learning environment so students understand their context, interact effectively, and see their value as customised learning companions.
In this post, we share four strategies to help introduce and integrate Cogniti in your teaching so that students understand their context, interact effectively, and see their value as customised learning companions.
Collection: Teaching with Custom AI Chatbots — from teaching.virginia.edu; via Derek Bruff The default behaviors of popular AI chatbots don’t always align with our teaching goals. This collection explores approaches to designing AI chatbots for particular pedagogical purposes.
While it’s true that Nano Banana generates better infographics than other AI models, the conversation has so far massively under-sold what’s actually different and valuable about this tool for those of us who design learning experiences.
What this means for our workflow:
Instead of the traditional “commission ? wait ? tweak ? approve ? repeat” cycle, Nano Banana enables an iterative, rapid-cycle design process where you can:
Sketch an idea and see it refined in minutes.
Test multiple visual metaphors for the same concept without re-briefing a designer.
Build 10-image storyboards with perfect consistency by specifying the constraints once, not manually editing each frame.
Implement evidence-based strategies (contrasting cases, worked examples, observational learning) that are usually too labour-intensive to produce at scale.
This shift—from “image generation as decoration” to “image generation as instructional scaffolding”—is what makes Nano Banana uniquely useful for the 10 evidence-based strategies below.
Three years ago, we were impressed that a machine could write a poem about otters. Less than 1,000 days later, I am debating statistical methodology with an agent that built its own research environment. The era of the chatbot is turning into the era of the digital coworker. To be very clear, Gemini 3 isn’t perfect, and it still needs a manager who can guide and check it. But it suggests that “human in the loop” is evolving from “human who fixes AI mistakes” to “human who directs AI work.” And that may be the biggest change since the release of ChatGPT.
Results May Vary — from aiedusimplified.substack.com by Lance Eaton, PhD On Custom Instructions with GenAI Tools….
I’m sharing today about custom instructions and my use of them across several AI tools (paid versions of ChatGPT, Gemini, and Claude). I want to highlight what I’m doing, how it’s going, and solicit from readers to share in the comments some of their custom instructions that they find helpful.
I’ve been in a few conversations lately that remind me that not everyone knows about them, even some of the seasoned folks around GenAI and how you might set them up to better support your work. And, of course, they are, like all things GenAI, highly imperfect!
I’ll include and discuss each one below, but if you want to keep abreast of my custom instructions, I’ll be placing them here as I adjust and update them so folks can see the changes over time.
My take is this: in all of the anxiety lies a crucial and long-overdue opportunity to deliver better learning experiences. Precisely because Atlas perceives the same context in the same moment as you, it can transform learning into a process aligned with core neuro-scientific principles—including active retrieval, guided attention, adaptive feedback and context-dependent memory formation.
Perhaps in Atlas we have a browser that for the first time isn’t just a portal to information, but one which can become a co-participant in active cognitive engagement—enabling iterative practice, reflective thinking, and real-time scaffolding as you move through challenges and ideas online.
With this in mind, I put together 10 use cases for Atlas for you to try for yourself.
…
6. Retrieval Practice
What: Pulling information from memory drives retention better than re-reading. Why: Practice testing delivers medium-to-large effects (Adesope et al., 2017). Try: Open a document with your previous notes. Ask Atlas for a mixed activity set: “Quiz me on the Krebs cycle—give me a near-miss, high-stretch MCQ, then a fill-in-the-blank, then ask me to explain it to a teen.” Atlas uses its browser memory to generate targeted questions from your actual study materials, supporting spaced, varied retrieval.
From DSC: A quick comment. I appreciate these ideas and approaches from Katarzyna and Rita. I do think that someone is going to want to be sure that the AI models/platforms/tools are given up-to-date information and updated instructions — i.e., any new procedures, steps to take, etc. Perhaps I’m missing the boat here, but an internal AI platform is going to need to have access to up-to-date information and instructions.
