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.
Why does it matter?
AI voice cloning has already flooded the internet with unauthorized imitations, blurring legal and ethical lines. By offering a dynamic, rights-secured platform, ElevenLabs aims to legitimize the booming AI voice industry and enable transparent, collaborative commercialization of iconic IP. .
. Data released by OpenAI in September from an internal study of queries sent to ChatGPT showed that most are for personal use, not work.
Emotional conversations were also common in the conversations analyzed by The Post, and users often shared highly personal details about their lives. In some chats, the AI tool could be seen adapting to match a user’s viewpoint, creating a kind of personalized echo chamber in which ChatGPT endorsed falsehoods and conspiracy theories.
Lee Rainie, director of the Imagining the Digital Future Center at Elon University, said his own research has suggested ChatGPT’s design encourages people to form emotional attachments with the chatbot. “The optimization and incentives towards intimacy are very clear,” he said. “ChatGPT is trained to further or deepen the relationship.”
Per The Rundown:OpenAI just shared its view on AI progress, predicting systems will soon become smart enough to make discoveries and calling for global coordination on safety, oversight, and resilience as the technology nears superintelligent territory.
The details:
OpenAI said current AI systems already outperform top humans in complex intellectual tasks and are “80% of the way to an AI researcher.”
The company expects AI will make small scientific discoveries by 2026 and more significant breakthroughs by 2028, as intelligence costs fall 40x per year.
For superintelligent AI, OAI said work with governments and safety agencies will be essential to mitigate risks like bioterrorism or runaway self-improvement.
It also called for safety standards among top labs, a resilience ecosystem like cybersecurity, and ongoing tracking of AI’s real impact to inform public policy.
Why it matters: While the timeline remains unclear, OAI’s message shows that the world should start bracing for superintelligent AI with coordinated safety. The company is betting that collective safeguards will be the only way to manage risk from the next era of intelligence, which may diffuse in ways humanity has never seen before.
Which linked to:
AI progress and recommendations — from openai.com AI is unlocking new knowledge and capabilities. Our responsibility is to guide that power toward broad, lasting benefit.
From DSC: I hate to say this, but it seems like there is growing concern amongst those who have pushed very hard to release as much AI as possible — they are NOW worried. They NOW step back and see that there are many reasons to worry about how these technologies can be negatively used.
Where was this level of concern before (while they were racing ahead at 180 mph)? Surely, numerous and knowledgeable people inside those organizations warned them about the destructive/downside of these technologies. But their warnings were pretty much blown off (at least from my limited perspective).
Most organizations are still in the experimentation or piloting phase: Nearly two-thirds of respondents say their organizations have not yet begun scaling AI across the enterprise.
High curiosity in AI agents: Sixty-two percent of survey respondents say their organizations are at least experimenting with AI agents.
Positive leading indicators on impact of AI: Respondents report use-case-level cost and revenue benefits, and 64 percent say that AI is enabling their innovation. However, just 39 percent report EBIT impact at the enterprise level.
High performers use AI to drive growth, innovation, and cost: Eighty percent of respondents say their companies set efficiency as an objective of their AI initiatives, but the companies seeing the most value from AI often set growth or innovation as additional objectives.
Redesigning workflows is a key success factor: Half of those AI high performers intend to use AI to transform their businesses, and most are redesigning workflows.
Differing perspectives on employment impact: Respondents vary in their expectations of AI’s impact on the overall workforce size of their organizations in the coming year: 32 percent expect decreases, 43 percent no change, and 13 percent increases.
Spatial intelligence is the next frontier in AI, demanding powerful world models to realize its full potential. World models should reconstruct, generate, and simulate 3D worlds; and allow both humans and agents to interact with them. Spatially intelligent world models will transform a wide variety of industries over the coming years.
Two months ago we shared a preview of Marble, our World Model that creates 3D worlds from image or text prompts. Since then, Marble has been available to an early set of beta users to create 3D worlds for themselves.
Today we are making Marble, a first-in-class generative multimodal world model, generally available for anyone to use. We have also drastically expanded Marble’s capabilities, and are excited to highlight them here:
I just completed nearly 60,000 miles of travel across Europe, Asia, and the Middle East meeting with hundred of companies to discuss their AI strategies. While every company’s maturity is different, one thing is clear: AI as a business tool has arrived: it’s real and the use-cases are growing.
A new survey by Wharton shows that 46% of business leaders use Gen AI daily and 80% use it weekly. And among these users, 72% are measuring ROI and 74% report a positive return. HR, by the way, is the #3 department in use cases, only slightly behind IT and Finance.
What are companies getting out of all this? Productivity. The #1 use case, by far, is what we call “stage 1” usage – individual productivity.
.
From DSC: Josh writes: “Many of our large clients are now implementing AI-native learning systems and seeing 30-40% reduction in staff with vast improvements in workforce enablement.”
While I get the appeal (and ROI) from management’s and shareholders’ perspective, this represents a growing concern for employment and people’s ability to earn a living.
And while I highly respect Josh and his work through the years, I disagree that we’re over the problems with AI and how people are using it:
Two years ago the NYT was trying to frighten us with stories of AI acting as a romance partner. Well those stories are over, and thanks to a $Trillion (literally) of capital investment in infrastructure, engineering, and power plants, this stuff is reasonably safe.
Those stories are just beginning…they’re not close to being over.
So let’s imagine a world where there’s no separation between learning and assessment: it’s ongoing. There’s always assessment, always learning, and they’re tied together. Then we can ask: what is the role of the human in that world? What is it that AI can’t do?
…
Imagine something like that in higher ed. There could be tutoring or skill-based work happening outside of class, and then relationship-based work happening inside of class, whether online, in person, or some hybrid mix.
The aspects of learning that don’t require relational context could be handled by AI, while the human parts remain intact. For example, I teach strategy and strategic management. I teach people how to talk with one another about the operation and function of a business. I can help students learn to be open to new ideas, recognize when someone pushes back out of fear of losing power, or draw from my own experience in leading a business and making future-oriented decisions.
But the technical parts such as the frameworks like SWOT analysis, the mechanics of comparing alternative viewpoints in a boardroom—those could be managed through simulations or reports that receive immediate feedback from AI. The relational aspects, the human mentoring, would still happen with me as their instructor.
What if your workspace helped you level up your career? Coworking spaces are becoming learning hubs where skills grow, ideas connect, and real-world education fits seamlessly into the workday.
Continuous learning has become a cornerstone of professional longevity, and flexible workspaces already encourage it through workshops, talks, and mentoring. Their true potential, however, may lie in becoming centers of industry-focused education that help professionals stay adaptable in a rapidly changing world of work. .
.
What if forward-thinking workspaces and coworking centers became hubs of lifelong learning, integrating job-relevant training with accessible, real-world education?
For coworking operators, this raises important questions: Which types of learning thrive best in these environments, and how much do the design and layout of a space influence how people learn?
By exploring these questions and combining innovative programs with cutting-edge technology aligned to the future workforce, could coworking spaces ultimately become the classrooms of tomorrow?
Law firm leaders should evaluate their legal technology and decide if they are truly helping legal work or causing a disconnect between human and AI contributions.
75% of firms now rely on cloud platforms for everything from document storage to client collaboration.
The rise of virtual law firms and remote work is reshaping the profession’s culture. Hybrid and remote-first models, supported by cloud and collaboration tools, are growing.
Are we truly innovating, or just rearranging the furniture? That’s the question every law firm leader should be asking as the legal technology landscape shifts beneath our feet. There are many different thoughts and opinions on how the legal technology landscape will evolve in the coming years, particularly regarding the pace of generative AI-driven changes and the magnitude of these changes.
To try to answer the question posed above, we looked at six recently published technology trends reports from influential entities in the legal technology arena: the American Bar Association, Clio, Wolters Kluwer, Lexis Nexis, Thomson Reuters, and NetDocuments.
When we compared these reports, we found them to be remarkably consistent. While the level of detail on some topics varied across the reports, they identified six trends that are reshaping the very core of legal practice. These trends are summarized in the following paragraphs.
The Big Four consulting and accounting firm is training its junior consultants to manage teams of AI agents — digital assistants capable of completing tasks without human input.
“We want juniors to become managers of agents,” Niale Cleobury, KPMG’s global AI workforce lead, told Business Insider in an interview.
KPMG plans to give new consulting recruits access to a catalog of AI agents capable of creating presentation slides, analyzing data, and conducting in-depth research, Cleobury said.
The goal is for these agents to perform much of the analytical and administrative work once assigned to junior consultants, allowing them to become more involved in strategic decisions.
From DSC: For a junior staff member to provide quality assurance in working with agents, an employee must know what they’re talking about in the first place. They must have expertise and relevant knowledge. Otherwise, how will they spot the hallucinations?
So the question is, how can businesses build such expertise in junior staff members while they are delegating things to an army of agents? This question applies to the next posting below as well. Having agents report to you is all well and good — IF you know when the agents are producing helpful/accurate information and when they got things all wrong.
Summary: A fundamental shift is making knowledge workers “AI managers.” The most valuable employees will direct intelligent AI agents, which requires new competencies: delegation, quality assurance and workflow orchestration across multiple agents. Companies must bridge the training gap to enable this move from simple software use to strategic collaboration with intelligent, yet imperfect, systems.
The shift is happening subtly, but it’s happening. Workers are learning to prompt agents, navigate AI capabilities, understand failure modes and hand off complex tasks to AI. And if they haven’t started yet, they probably will: A new study from IDC and Salesforce found that 72 percent of CEOs think most employees will have an AI agent reporting to them within five years. This isn’t about using a new kind of software tool — it’s about directing intelligent systems that can reason, search, analyze and create.
Soon, the most valuable employees won’t just know how to use AI; they’ll know how to manage it. And that requires a fundamentally different skill set than anything we’ve taught in the workplace before.
AI Agents Can’t Actually Do Your Job (Yet)—New Benchmark Reveals The Gap
DEEP DIVE: AI can make you faster at your job, but can only do 2-3% of jobs by itself.
The hype: AI agents will automate entire workflows! Replace freelancers! Handle complex tasks end-to-end!
The reality: a measly 2-3% completion rate.
See, Scale AI and CAIS just released the Remote Labor Index (paper), a benchmark where AI agents attempted real freelance tasks. The best-performing model earned just $1,810 out of $143,991 in available work, and yes, finishing only 2-3% of jobs.
From DSC: One of my sisters shared this piece with me. She is very concerned about our society’s use of technology — whether it relates to our youth’s use of social media or the relentless pressure to be first in all things AI. As she was a teacher (at the middle school level) for 37 years, I greatly appreciate her viewpoints. She keeps me grounded in some of the negatives of technology. It’s important for us to listen to each other.
Nvidia has officially become the first company in history to cross the $5 trillion market cap, cementing its position as the undisputed leader of the AI era. Just three months ago, the chipmaker hit $4 trillion; it’s already added another trillion since.
Nvidia market cap milestones:
Jan 2020: $144 billion
May 2023: $1 trillion
Feb 2024: $2 trillion
Jun 2024: $3 trillion
Jul 2025: $4 trillion
Oct 2025: $5 trillion
The above posting linked to:
Nvidia becomes first public company worth $5 trillion — from techcrunch.com by Ivan Mehta The biggest beneficiary of the ongoing AI boom, Nvidia has become the first public company to pass the $5 trillion market cap milestone.
This blog explores how custom AI development accelerates the evolution from static AI to dynamic learning agents and why this transformation is critical for driving innovation, efficiency, and competitive advantage.
… Dynamic Learning Agents: The Next Generation Dynamic learning agents, sometimes referred to as adaptive or agentic AI, represent a leap forward. They combine continuous learning, autonomous action, and context-aware adaptability.
Custom AI development plays a crucial role here: it ensures that these agents are designed specifically for an enterprise’s unique needs rather than relying on generic, one-size-fits-all AI platforms. Tailored dynamic agents can:
Continuously learn from incoming data streams
Make autonomous, goal-directed decisions aligned with business objectives
Adapt behavior in real time based on context and feedback
Collaborate with other AI agents and human teams to solve complex challenges
The result is an AI ecosystem that evolves with the business, providing sustained competitive advantage.
Perception: The Foundation of Intelligent Agents Perception is the first step in building AI agents. It involves capturing and interpreting data from multiple modalities, including text, images, audio, and structured inputs. A multimodal AI agent relies on this comprehensive understanding to make informed decisions.
For example, in healthcare, an AI agent may process electronic health records (text), MRI scans (vision), and patient audio consultations (speech) to build a complete understanding of a patient’s condition. Similarly, in retail, AI agents can analyze purchase histories (structured data), product images (vision), and customer reviews (text) to inform recommendations and marketing strategies.
Effective perception ensures that AI agents have contextual awareness, which is essential for accurate reasoning and appropriate action.
Your New ChatGPT Guide— from wondertools.substack.com by Jeremy Caplan and The PyCoach 25 AI Tips & Tricks from a guest expert
ChatGPT can make you more productive or dumber. An MIT study found that while AI can significantly boost productivity, it may also weaken your critical thinking. Use it as an assistant, not a substitute for your brain.
If you’re a student, use study mode in ChatGPT, Gemini, or Claude. When this feature is enabled, the chatbots will guide you through problems rather than just giving full answers, so you’ll be doing the critical thinking.
ChatGPT and other chatbots can confidently make stuff up (aka AI hallucinations). If you suspect something isn’t right, double-check its answers.
NotebookLM hallucinates less than most AI tools, but it requires you to upload sources (PDFs, audio, video) and won’t answer questions beyond those materials. That said, it’s great for students and anyone with materials to upload.
Probably the most underrated AI feature is deep research. It automates web searching for you and returns a fully cited report with minimal hallucinations in five to 30 minutes. It’s available in ChatGPT, Perplexity, and Gemini, so give it a try.
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.
From DSC: Stephen has some solid reflections and asks some excellent questions in this posting, including:
The question is: how do we optimize an AI to support learning? Will one model be enough? Or do we need different models for different learners in different scenarios?
A More Human University: The Role of AI in Learning — from er.educause.edu by Robert Placido Far from heralding the collapse of higher education, artificial intelligence offers a transformative opportunity to scale meaningful, individualized learning experiences across diverse classrooms.
The narrative surrounding artificial intelligence (AI) in higher education is often grim. We hear dire predictions of an “impending collapse,” fueled by fears of rampant cheating, the erosion of critical thinking, and the obsolescence of the human educator.Footnote1 This dystopian view, however, is a failure of imagination. It mistakes the death rattle of an outdated pedagogical model for the death of learning itself. The truth is far more hopeful: AI is not an asteroid coming for higher education. It is a catalyst that can finally empower us to solve our oldest, most intractable problem: the inability to scale deep, engaged, and truly personalized learning.
Increasing the rate of scientific progress is a core part of Anthropic’s public benefit mission.
We are focused on building the tools to allow researchers to make new discoveries – and eventually, to allow AI models to make these discoveries autonomously.
Until recently, scientists typically used Claude for individual tasks, like writing code for statistical analysis or summarizing papers. Pharmaceutical companies and others in industry also use it for tasks across the rest of their business, like sales, to fund new research. Now, our goal is to make Claude capable of supporting the entire process, from early discovery through to translation and commercialization.
To do this, we’re rolling out several improvements that aim to make Claude a better partner for those who work in the life sciences, including researchers, clinical coordinators, and regulatory affairs managers.
AI as an access tool for neurodiverse and international staff— from timeshighereducation.com by Vanessa Mar-Molinero Used transparently and ethically, GenAI can level the playing field and lower the cognitive load of repetitive tasks for admin staff, student support and teachers
Where AI helps without cutting academic corners When framed as accessibility and quality enhancement, AI can support staff to complete standard tasks with less friction. However, while it supports clarity, consistency and inclusion, generative AI (GenAI) does not replace disciplinary expertise, ethical judgement or the teacher–student relationship. These are ways it can be put to effective use:
The Sleep of Liberal Arts Produces AI — from aiedusimplified.substack.com by Lance Eaton, Ph.D. A keynote at the AI and the Liberal Arts Symposium Conference
This past weekend, I had the honor to be the keynote speaker at a really fantstistic conferece, AI and the Liberal Arts Symposium at Connecticut College. I had shared a bit about this before with my interview with Lori Looney. It was an incredible conference, thoughtfully composed with a lot of things to chew on and think about.
It was also an entirely brand new talk in a slightly different context from many of my other talks and workshops. It was something I had to build entirely from the ground up. It reminded me in some ways of last year’s “What If GenAI Is a Nothingburger”.
It was a real challenge and one I’ve been working on and off for months, trying to figure out the right balance. It’s a work I feel proud of because of the balancing act I try to navigate. So, as always, it’s here for others to read and engage with. And, of course, here is the slide deck as well (with CC license).
The Most Innovative Law Schools (2025) — from abovethelaw.com by Staci Zaretsky Forget dusty casebooks — today’s leaders in legal education are using AI, design thinking, and real-world labs to reinvent how law is taught.
“[F]rom AI labs and interdisciplinary centers to data-driven reform and bold new approaches to design and client service,” according to National Jurist’s preLaw Magazine, these are the law schools that “exemplify innovation in action.”
North Carolina Central University School of Law
Suffolk University Law School
UC Berkeley School of Law
Nova Southeastern University Shepard Broad College of Law
Northeastern University School of Law
Maurice A. Deane School of Law at Hofstra University
Seattle University School of Law
Case Western Reserve University School of Law
University of Miami School of Law
Benjamin N. Cardozo School of Law at Yeshiva University
Vanderbilt University Law School
Southwestern Law School
Click here to read short summaries of why each school made this year’s list of top innovators.
Newton delivered what may have been the most consequential keynote in the company’s history and one that signals a shift by Clio from a traditional practice management provider to a comprehensive platform that essentially does everything for the business and practice of law.
…
Clio also earlier this year acquired vLex, the heavy-duty AI legal research player. The acquisition is pending regulatory approval. It is the vLex acquisition that is powering the Clio transformation that Newton described in his keynote.
vLex has a huge amount of legal data in its wheelhouse to power sophisticated legal AI research. On top of this data, vLex developed Vincent, a powerful AI tool to work with this data and enable all sorts of actions and work.
This means a couple of things. First, by acquiring vLex, Clio can now offer its customers AI legal research tools. Clio customers will no longer have to go one place for its practice management needs and a second place for its substantive legal work, like research. It makes what Clio can provide much more comprehensive and all inclusive.
Artificial intelligence will give solos and small firms “a huge advantage,” according to one legal tech consultant.
In this episode of “Adventures in Legal Tech,” host Jared Correia speaks with Ernie Svenson — aka “Ernie the Attorney” — about the psychology behind resistance to change, how law firms are positioning their AI use, the power of technology for business development, and more.
Legal ops experts can categorize legal AI platforms and software by the ability to streamline key tasks such as legal research, document processing or analysis, and drafting.
The trustworthiness and accuracy of AI hinge on the quality of its underlying data; solutions like CoCounsel Legal are grounded in authoritative, expert-verified content from Westlaw and Practical Law, unlike providers that may rely on siloed or less reliable databases.
When evaluating legal software, firms should use a framework that assesses critical factors such as integration with existing tech stacks, security, scalability, user adoption, and vendor reputation.
The Sandra Day O’Connor College of Law at Arizona State University appointed Sean Harrington as director of the newly established AI and Legal Tech Studio, a key milestone in ASU Law’s bold initiative to reimagine legal education for the artificial intelligence era. ASU, ranked No. 1 in innovation for the 11th consecutive year, drives AI solutions that enhance teaching, enrich student training and facilitate digital transformation.
The sixth annual American Legal Technology Awards were presented on Wednesday, October 15th, at Suffolk University Law School (Boston), recognizing winners across ten categories. There were 211 nominees who were evaluated by 27 judges.
“Future of Professionals Report” analysis: Why AI will flip law firm economics — from thomsonreuters.com by Ragunath Ramanathan AI forces a reinvention of law firm billing models, the market will reward those firms that price by outcome, guarantee efficiency, and are transparent. The question then isn’t whether to change — it’s whether firms will stand on the sidelines or lead
Key insights:
Efficiency and cost savings are expected— AI is significantly increasing efficiency and reducing costs in the legal industry, with each lawyer expecting to save 190 work-hours per year by leveraging AI, resulting in approximately $20 billion worth of work-savings in the US alone.
Challenges to the billable hour model— The traditional billable hour model is being challenged by AI advancements, as lawyers are now able to complete tasks more efficiently and quickly, leading some law firms to explore alternative pricing models that reflect the value delivered rather than the time spent.
Opportunities for smaller law firms— AI presents unique opportunities for smaller law firms to differentiate themselves and compete with larger firms, as AI solutions allow smaller firms to access advanced technology without significant investment and deliver innovative pricing models.
The legal industry is undergoing a significant transformation that’s being driven by the rapid adoption of AI — a shift that is poised to redefine traditional practices, particularly the billable hour model, a cornerstone of law firm operations.
…
Not surprisingly, AI is anticipated to have the biggest impact on the legal industry over the next five years, with 80% of law firm survey respondents to Thomson Reuters recently published 2025 Future of Professionals report saying that they expect AI to fundamentally alter how they conduct business, especially around how law firms price, staff, and deliver legal work to their clients.
Experience AI: A new architecture of learning
Experience AI represents a new architecture for learning — one that prioritizes continuity, agency and deep personalization. It fuses three dimensions into a new category of co-intelligent systems:
Agentic AI that evolves with the learner, not just serves them
Persona-based AI that adapts to individual goals, identities and motivations
Multimodal AI that engages across text, voice, video, simulation and interaction
Experience AI brings learning into context. It powers personalized, problem-based journeys where students explore ideas, reflect on progress and co-create meaning — with both human and machine collaborators.
In short, it’s been a monumental 12 months for AI. Our eighth annual report is the most comprehensive it’s ever been, covering what you need to know about research, industry, politics, and safety – along with our first State of AI Usage Survey of 1,200 practitioners.
