Deadbots. Many Use Cases, Including Advertising
September 2, 2025
No AI. Just a dinobaby working the old-fashioned way.
I like the idea of deadbots, a concept explained by the ever-authoritative NPR in “AI Deadbots Are Persuasive — and Researchers Say, They’re Primed for Monetization.” The write up reports in what I imagine as a resonant, somewhat breathy voice:
AI avatars of deceased people – or “deadbots” – are showing up in new and unexpected contexts, including ones where they have the power to persuade.
Here’s a passage I thought was interesting:
Researchers are now warning that commercial use is the next frontier for deadbots. “Of course it will be monetized,” said Lindenwood University AI researcher James Hutson. Hutson co-authored several studies about deadbots, including one exploring the ethics of using AI to reanimate the dead. Hutson’s work, along with other recent studies such as one from Cambridge University, which explores the likelihood of companies using deadbots to advertise products to users, point to the potential harms of such uses. “The problem is if it is perceived as exploitative, right?” Hutson said.
Not surprisingly, some sticks in the mud see a downside to deadbots:
Quinn [a wizard a Authetic Interactions Inc.] said companies are going to try to make as much money out of AI avatars of both the dead and the living as possible, and he acknowledges there could be some bad actors. “Companies are already testing things out internally for these use cases,” Quinn said, with reference to such uses cases as endorsements featuring living celebrities created with generative AI that people can interactive with. “We just haven’t seen a lot of the implementations yet.”
I wonder if any philosophical types will consider how an interaction with a dead person’s avatar can be an “authetic interaction.”
I started thinking of deadbots I would enjoy coming to life on my digital devices; for example:
- My first boss at a blue chip consulting firm who encouraged rumors that his previous wives accidently met with boating accidents
- My high school English teacher who took me to the assistant principal’s office for writing a poem about the spirit of nature who looked to me like a Playboy bunny
- The union steward who told me that I was working too fast and making other workers look like they were not working hard
- The airline professional who told me our flight would be delayed when a passenger died during push back from the gate. (The fellow was sitting next to me. Airport food did it I think.)
- The owner of an enterprise search company who insisted, “Our enterprise information retrieval puts all your company’s information at an employee’s fingertips.”
You may have other ideas for deadbots. How would you monetize a deadbot, Google- and Meta-type companies? Will Hollywood do deadbot motion pictures? (I know the answer to that question.)
Stephen E Arnold, September 2, 2025
AI Will Not Have a Negative Impact on Jobs. Knock Off the Negativity Now
September 2, 2025
No AI. Just a dinobaby working the old-fashioned way.
The word from Goldman Sachs is parental and well it should be. After all, Goldman Sachs is the big dog. PC Week’s story “Goldman Sachs: AI’s Job Hit Will Be Brief as Productivity Rises” makes this crystal clear or almost. In an era of PR and smart software, I am never sure who is creating what.
The write up says:
AI will cause significant, but ultimately temporary, disruption. The headline figure from the report is that widespread adoption of AI could displace 6-7% of the US workforce. While that number sounds alarming, the firm’s economists, Joseph Briggs and Sarah Dong, argue against the narrative of a permanent “jobpocalypse.” They remain “skeptical that AI will lead to large employment reductions over the next decade.”
Knock of the complaining already. College graduates with zero job offers. Just do the van life thing for a decade or become an influencer.
The write up explains history just like the good old days:
“Predictions that technology will reduce the need for human labor have a long history but a poor track record,” they write. The report highlights a stunning fact: Approximately 60% of US workers today are employed in occupations that didn’t even exist in 1940. This suggests that over 85% of all employment growth in the last 80 years has been fueled by the creation of new jobs driven by new technologies. From the steam engine to the internet, innovation has consistently eliminated some roles while creating entirely new industries and professions.
Technology and brilliant management like that at Goldman Sachs makes the economy hum along. And the write up proves it, and I quote:
Goldman Sachs expects AI to follow this pattern.
For those TikTok- and YouTube-type videos revealing that jobs are hard to obtain or the fathers whining about sending 200 job applications each month for six months, knock it off. The sun will come up tomorrow. The financial engines will churn and charge a service fee, of course. The flowers will bloom because that baloney about global warming is dead wrong. The birds will sing (well, maybe not in Manhattan) but elsewhere because windmills creating power are going to be shut down so the birds won’t be decapitated any more.
Everything is great. Goldman Sachs says this. In Goldman we trust or is it Goldman wants your trust… fund that is.
Stephen E Arnold, September 2, 2025
More about AI and Peasants from a Xoogler Too
September 1, 2025
A former Googler predicts a rough ride ahead for workers. And would-be workers. Yahoo News shares “Ex-Google Exec’s Shocking Warning: AI Will Create 15 Years of ‘Hell’—Starting Sooner than We Think.” Only 15 years? Seems optimistic. Mo Gawdat issued his prophesy on the “Diary of a CEO” podcast. He expects “the end of white-collar work” to begin by the end of this decade. Indeed, the job losses have already begun. But the cascading effects could go well beyond high unemployment. Reporter Ariel Zilber writes:
“Without proper government oversight, AI technology will channel unprecedented wealth and influence to those who own or control these systems, while leaving millions of workers struggling to find their place in the new economy, according to Gawdat. Beyond economic concerns, Gawdat anticipates serious social consequences from this rapid transformation. Gawdat said AI will trigger significant ‘social unrest’ as people grapple with losing their livelihoods and sense of purpose — resulting in rising rates of mental health problems, increased loneliness and deepening social divisions. ‘Unless you’re in the top 0.1%, you’re a peasant,’ Gawdat said. ‘There is no middle class.’”
That is ominous. But, to hear Gawdat tell it, there is a bright future on the other side of those hellish 15 years. He believes those who survive past 2040 can look forward to a “utopian” era free from tedious, mundane tasks. This will free us up to focus on “love, community, and spiritual development.” Sure. But to get there, he warns, we must take certain steps:
“Gawdat said that it is incumbent on governments, individuals and businesses to take proactive measures such as the adoption of universal basic income to help people navigate the transition. ‘We are headed into a short-term dystopia, but we can still decide what comes after that,’ Gawdat told the podcast, emphasizing that the future remains malleable based on choices society makes today. He argued that outcomes will depend heavily on decisions regarding regulation, equitable access to technology, and what he calls the ‘moral programming’ of AI algorithms.”
We are sure government and Big Tech will get right on that. Totally doable in our current political and business climates. Meanwhile, Mo Gawdat is working on an “AI love coach.” I am not sure Mr. Gawdat is connected to the bureaucratic and management ethos of 2025. Is that why he is a Xoogler?
Cynthia Murrell, September 1, 2025
Faux Boeuf Delivers Zero Calories Plus a Non-Human Toxin
August 29, 2025
No AI. Just a dinobaby working the old-fashioned way.
That sizzling rib AI called boeuf à la Margaux Blanchard is a treat. I learned about this recipe for creating filling, substantive, calorie laden content in “Wired and Business Insider Remove Articles by AI-Generated Freelancer.” I can visualize the meeting in which the decision was taken to hire Margaux Blanchard. I can also run in my mental VHS, the meeting when the issue was discovered. In my version, the group agreed to blame it on a contractor and the lousy job human resource professionals do these days.
What’s the “real” story? Let go to the Guardian write up:
On Thursday [August 22, 2025], Press Gazette reported that at least six publications, including Wired and Business Insider, have removed articles from their websites in recent months after it was discovered that the stories – written under the name of Margaux Blanchard – were AI-generated.
I frequently use the phrase “ordained officiant” in my dinobaby musings. Doesn’t everyone with some journalism experience?
The write u p said:
Wired’s management acknowledged the faux pas, saying: “If anyone should be able to catch an AI scammer, it’s Wired. In fact we do, all the time … Unfortunately, one got through. We made errors here: This story did not go through a proper fact-check process or get a top edit from a more senior editor … We acted quickly once we discovered the ruse, and we’ve taken steps to ensure this doesn’t happen again. In this new era, every newsroom should be prepared to do the same.”
Yeah, unfortunately and quickly. Yeah.
I liked this paragraph in the story:
This incident of false AI-generated reporting follows a May error when the Chicago Sun-Times’ Sunday paper ran a syndicated section with a fake reading list created by AI. Marco Buscaglia, a journalist who was working for King Features Syndicate, turned to AI to help generate the list, saying: “Stupidly, and 100% on me, I just kind of republished this list that [an AI program] spit out … Usually, it’s something I wouldn’t do … Even if I’m not writing something, I’m at least making sure that I correctly source it and vet it and make sure it’s all legitimate. And I definitely failed in that task.” Meanwhile, in June, the Utah court of appeals sanctioned a lawyer after he was discovered to have used ChatGPT for a filing he made in which he referenced a nonexistent court case.
Hey, that AI is great. It builds trust. It is intellectually satisfying just like some time in the kitchen with Margot Blanchard, a hot laptop, and some spicy prompts. Yum yum yum.
Stephen E Arnold, August 29, 2025
Google Uses a Blue Light Special for the US Government (Sorry K-Meta You Lose)
August 27, 2025
No AI. Just a dinobaby working the old-fashioned way.
I read an interesting news item in Artificial Intelligence News, a publication unknown to me. Like most of the AI information I read online I believe every single word. AI radiates accuracy, trust, and factual information. Let’s treat this “real” news story as actual factual. To process the information, you will want to reflect on the sales tactics behind Filene’s Basement, K-Mart’s blue light specials, and the ShamWow guy.
“The US Federal Government Secures a Massive Google Gemini AI Deal at $0.47 per Agency” reports:
Google Gemini will soon power federal operations across the United States government following a sweeping new agreement between the General Services Administration (GSA) and Google that delivers comprehensive AI capabilities at unprecedented pricing.
I regret I don’t have Microsoft government sales professional or a Palantir forward deployed engineer to call and get their view of this deal. Oh, well, that’s what happens when one gets old. (Remember. For a LinkedIn audience NEVER reveal your age. Okay, too bad LinkedIn, I am 81.)
It so happens I was involved in Year 2000 in some meetings at which Google pitched its search-and-retrieval system for US government wide search. For a number of reasons, the Google did not win that procurement bake off. It took a formal protest and some more meetings to explain the concept of conforming to a Statement of Work and the bid analysis process used by the US government 25 years ago. Google took it on the snout.
Not this time.
By golly, Google figured out how to deal with RFPs, SOWs, the Q&A process, and the pricing dance. The write up says:
The “Gemini for Government” offering, announced by GSA, represents one of the most significant government AI procurement deals to date. Under the OneGov agreement extending through 2026, federal agencies will gain access to Google’s full artificial intelligence stack for just US$0.47 per agency—a pricing structure that industry observers note is remarkably aggressive for enterprise-level AI services.
What does the US government receive? According to the write up:
Google CEO Sundar Pichai characterized the partnership as building on existing relationships: “Building on our Workspace offer for federal employees, ‘Gemini for Government’ gives federal agencies access to our full stack approach to AI innovation, including tools like NotebookLM and Veo powered by our latest models and our secure cloud infrastructure.”
Yo, Microsoft. Yo, Palantir. Are you paying attention? This explanation suggests that a clever government professional can do what your firms do. But — get this — at a price that may be “unsustainable.” (Of course, I know that em dashes signal smart software. Believe me. I use em dashes all by myself. No AI needed.)
I also noted this statement in the write up:
The $0.47 per agency pricing model raises immediate concerns about market distortion and the sustainability of such aggressive government contracting. Industry analysts question whether this represents genuine cost efficiency or a loss-leader strategy designed to lock agencies into Google’s ecosystem before prices inevitably rise after 2026. Moreover, the deal’s sweeping scope—encompassing everything from basic productivity tools to custom AI agent development—may create dangerous vendor concentration risks. Should technical issues, security breaches, or contract disputes arise, the federal government could find itself heavily dependent on a single commercial provider for critical operational capabilities. The announcement notably lacks specific metrics for measuring success, implementation timelines, or safeguards against vendor lock-in—details that will ultimately determine whether this represents genuine modernization or expensive experimentation with taxpayer resources.
Several observations are warranted:
- Google has figured out that making AI too cheap to resist appeals to certain government procurement professionals. A deal is a deal, of course. Scope changes, engineering services, and government budget schedules may add some jerked chicken spice to the bargain meal.
- The existing government-wide incumbent types are probably going to be holding some meetings to discuss what “this deal” means to existing and new projects involving smart software.
- The budget issues about AI investments are significant. Adding more expense for what can be a very demanding client is likely to have a direct impact on advertisers who fund the Google fun bus. How much will that YouTube subscription go up? Would Google raise rates to fund this competitive strike at Microsoft and Palantir? Of course not, you silly goose.
I wish I were at liberty to share some of the Google-related outputs from the Year 2000 procurement. But, alas, I cannot. Let me close by saying, “Google has figured out some basics of dealing with the US government.” Hey, it only took a quarter century, not bad for an ageing Googzilla.
Stephen E Arnold, August 27, 2025
Think It. The * It * Becomes Real. Think Again?
August 27, 2025
No AI. Just a dinobaby working the old-fashioned way.
Fortune Magazine — once the gem for a now spinning-in-his-grave publisher —- posted “MIT Report: 95% of Generative AI Pilots at Companies Are Failing.” I take a skeptical view of MIT. Why? The esteemed university found Jeffrey Epstein a swell person.
The thrust of the story is that people stick smart software into an organization, allow it time to steep, cook up a use case, and find the result unpalatable. Research is useful. When it evokes a “Duh!”, I don’t get too excited.
But there was a phrase in the write up which caught my attention: Learning gap. AI or smart software is a “belief.” The idea of the next big thing creates an opportunity to move money. Flow, churn, motion — These are positive values in some business circles.
AI fits the bill. The technology demonstrates interesting capabilities. Use cases exist. Companies like Microsoft have put money into the idea. Moving money is proof that “something” is happening. And today that something is smart software. AI is the “it” for the next big thing.
Learning gap, however, is the issue. The hurdle is not Sam Altman’s fears about the end of humanity or his casual observation that trillions of dollars are needed to make AI progress. We have a learning gap.
But the driving vision for Internet era innovation is do something big, change the world, reinvent society. I think this idea goes back to the sales-oriented philosophy of visualizing a goal and aligning one’s actions to achieve that goal. I a fellow or persona named Napoleon Hill pulled together some ideas and crafted “Think and Grow Rich.” Today one just promotes the “next big thing,” gets some cash moving, and an innovation like smart software will revolutionize, remake, or redo the world.
The “it” seems to be stuck in the learning gap. Here’s the proof, and I quote:
But for 95% of companies in the dataset, generative AI implementation is falling short. The core issue? Not the quality of the AI models, but the “learning gap” for both tools and organizations. While executives often blame regulation or model performance, MIT’s research points to flawed enterprise integration. Generic tools like ChatGPT excel for individuals because of their flexibility, but they stall in enterprise use since they don’t learn from or adapt to workflows, Challapally explained. The data also reveals a misalignment in resource allocation. More than half of generative AI budgets are devoted to sales and marketing tools, yet MIT found the biggest ROI in back-office automation—eliminating business process outsourcing, cutting external agency costs, and streamlining operations.
Consider this question: What if smart software mostly works but makes humans uncomfortable in ways difficult for the user to articulate? What if humans lack the mental equipment to conceptualize what a smart system does? What if the smart software cannot answer certain user questions?
I find information about costs, failed use cases, hallucinations, and benefits plentiful. I don’t see much information about the “learning gap.” What causes a learning gap? Spell check makes sense. A click that produces a complete report on a complex topic is different. But in what way? What is the impact on the user?
I think the “learning gap” is a key phrase. I think there is money to be made in addressing it. I am not confident that visualizing a better AI is going to solve the problem which is similar to a bonfire of cash. The learning gap might be tough to fill with burning dollar bills.
Stephen E Arnold, August 27, 2025
Apple and Meta: The After Market Route
August 26, 2025
No AI. Just a dinobaby working the old-fashioned way.
Two big outfits are emulating the creative motif for an American television series titled “Pimp My Ride.” The show was hosted by rapper Xzibit, who has a new album called “Kingmaker” in the works. He became the “meme” of the television program with his signature phrase, “Yo, dawg, I heard you like.”
A DVD of season one, is available for sale at www.bol.com.
Each episode a “lucky person” would be approached and told that his or her vehicle would be given a make over. Some of the make overs were memorable. Examples included the “Yellow Shag Disaster,” which featured yellow paint and yellow shag carpeting. The team removed a rat living in the 1976 Pacer. Another was the “Drive In Theater Car.” It included a pop up champagne dispenser and a TV screen installed under the hood for a viewing experience when people gathered outside the vehicle.
The idea was to take something that mostly worked and then add-on extras. Did the approach work? It made Xzibit even more famous and it contributed the phrase “Yo, dawg, I heard you like” to the US popular culture between 2004 and 2007.
I think the “Pimp My Ride” concept has returned for Apple and Meta. Let me share my thoughts with you.
First, I noted that Bloomberg is exploring the use of Google Gemini AI to Power the long suffering Siri. You can read the paywalled story at this link. Apple knows that Google’s payments are worth real money. The idea of adding more Google and getting paid for the decision probably makes sense to the estimable Apple. Will the elephants mate and produce more money or will the grass get trampled. I don’t know. It will be interesting to see what the creative wizards at both companies produce. There is no date for the release of the first episode. I will be watching.
Second, the story presented in fragments on X.com appears at this X.com page. The key item of information is the alleged tie up between Meta and MidJourney:
Today we’re proud to announce a partnership with @midjourney , to license their aesthetic technology for our future models and products, bringing beauty to billions.
Meta, like Apple, is partnering with an AI success in the arts and crafts sector of smart software. The idea seems to focus on “aesthetic excellence.” How will these outfits enhance Meta. Here’s what the X.com comment offers:
To ensure Meta is able to deliver the best possible products for people it will require taking an all-of-the-above approach. This means world-class talent, ambitious compute roadmap, and working with the best players across the industry.
Will these add-one approaches to AI deliver something useful to millions or will the respective organizations produce the equivalent of the “Pimp My Ride” Hot Tub Limousine. This after-market confection added a hot tub filled with water to a limousine. The owner of the vehicle could relax in the hot tub while the driver ferried the proud owner to the bank.
I assume the creations of the Apple, Google, Meta, and MidJourney teams will be captured on video and distributed on TikTok-type services as well as billions of computing devices. My hope is that Xzibit is asked to host the roll outs for the newly redone services. I would buy a hat, a T shirt, and a poster for the “winner” of this new AI enhanced effort.
Yo, dawg, I heard you like AI, right?
Stephen E Arnold, August 26, 2025
Deal Breakers in Medical AI
August 26, 2025
No AI. Just a dinobaby working the old-fashioned way.
My newsfeed thing spit out a link to “Why Radiology AI Didn’t Work and What Comes Next.” I have zero interest in radiology. I don’t get too excited about smart software. So what did I do? Answer: I read the article. I was delighted to uncover a couple of points that, in my opinion, warrant capturing in my digital notebook.
The set up is that a wizard worked at a start up trying to get AI to make sense of the consistently fuzzy, murky, and baffling images cranked out by radiology gizmos. Tip: Follow the instructions and don’t wear certain items of jewelry. The start up fizzled. AI was part of the problem, but the Jaws-type sharp lurking in the murky image explains this type of AI implosion.
Let’s run though the points that struck me.
First, let’s look at this passage:
Unlike coding or mathematics, medicine rarely deals in absolutes. Clinical documentation, especially in radiology, is filled with hedge language — phrases like “cannot rule out,” “may represent,” or “follow-up recommended for correlation.” These aren’t careless ambiguities; they’re defensive signals, shaped by decades of legal precedent and diagnostic uncertainty.
Okay, lawyers play a significant role in establishing thought processes and normalizing ideas that appear to be purpose-built to vaporize like one of those nifty tattoo removing gadgets the smart system. I would have pegged insurance companies, then lawyers, but the write up directed my attention of the legal eagles’ role: Hedge language. Do I have disease X? The doctor responds, “Maybe, maybe not. Let’s wait 30 days and run more tests.” Fuzzy lingo, fuzzy images, perfect.
Second, the write up asks two questions:
- How do we improve model coverage at the tail without incurring prohibitive annotation costs?
- Can we combine automated systems with human-in-the-loop supervision to address the rare but dangerous edge cases?
The answers seem to be: You cannot afford to have humans do indexing and annotation. That’s why certain legal online services charge a lot for annotations. And, the second question, no, you cannot pull off automation with humans for events rarely covered in the training data. Why? Cost and finding enough humans who will do this work in a consistent way in a timely manner.
Here’s the third snippet:
Without direct billing mechanisms or CPT reimbursement codes, it was difficult to monetize the outcomes these tools enabled. Selling software alone meant capturing only a fraction of the value AI actually created. Ultimately, we were offering tools, not outcomes. And hospitals, rightly, were unwilling to pay for potential unless it came bundled with performance.
Finally, insurance procedures. Hospitals aren’t buying AI; they are buying ways to deliver “service” and “bill.” AI at this time does not sell what hospitals want to buy: A way to keep high rates and slash costs wherever possible.
Unlikely but perhaps some savvy AI outfit will create a system that can crack the issues the article identifies. Until then, no money, no AI.
Stephen E Arnold, August 26, 2025
Learning Is Hard Work: AI Is Not Part of My Game Plan
August 25, 2025
No AI. Just a dinobaby working the old-fashioned way.
Dinobaby here—a lifetime of unusual education packed into a single childhood. I kicked off in a traditional Illinois kindergarten, then traded finger painting for experimental learning at a “new-idea” grade school in Maryland after a family move near DC. Soon, Brazil called: I landed in Campinas, but with zero English spoken, I lasted a month. Fifth through seventh grade became a solo mission—Calvert Course worksheets, a jungle missionary who mailed my work to Baltimore, and eventually, after the tutor died, pure self-guided study from thousands of miles away. I aced my assignments, but no one in Maryland had any idea of my world. My Portuguese tutor mixed French and German with local lingo; ironically, her English rocketed while my Portuguese crawled.
Back in the States, I dove into “advanced” classes and spent a high school semester at the University of Illinois—mainly reading, testing, and reading. A scholarship sent me to Bradley, a few weeks removed from a basketball cheating inquiry. A professor hooked me on coding in the library, building Latin sermon indexes using the school’s IBM. That led to a Duquesne fellowship; then the University of Arkansas wanted me for their PhD program. But I returned to Illinois, wrote code for Milton texts instead of Latin under Arthur Barker’s mentorship, and gave talks that landed me a job offer. One conference center chat brought me to DC and into the nuclear division at Halliburton. That’s my wild educational ride.
Notice that it did not involve much traditional go-to-class activity. I have done okay despite my somewhat odd educational journey. Most important: No smart software.
Now why did I provide this bit of biographical trivia? I read “AI in the Classroom Is Important for Real-World Skills, College Professors Say.” I did not have access to “regular” school through grade school, high school, and college. I am not sure how many high school students took classes at the U of I when they were 15 years old, but that experience was not typical among my high school class.
I did start working with computers and software in 1962, but there wasn’t much smart software floating around then. The trick for me has been my ability to read quickly, recognize what’s important, and remember information. Again there was no AI. Today, as I finish my Telegram Labyrinth monograph, AI has not been of any importance. Most of the source material is in Russian language documents. The English information is not thoroughly indexed by Telegram nor by the Web search engines. The LLM content suckers are not doing too much with information outside the English speaking world. Maybe China is pushing forward, but my tests with Chinese language Web search engines did not provide much, if any, information my team and I already had reviewed.
Obviously I don’t think AI is something that fits into my “real world skills.” The write up says:
“If integrated well, AI in the classroom can strengthen the fit between what students learn and what students will see in the workforce and world around them,” argued Victor Lee, associate professor at Stanford’s Graduate School of Education. GenAI companies are certainly doing their part to lure students into using their tools by offering new learning and essay-writing features. Google has gone so far as to offer Gemini free for one year, and OpenAI late last month introduced “Study Mode” to help students “work through problems step by step instead of just getting an answer,” the company said in a blog post.
Maybe.
My personal approach to learning involves libraries, for fee online databases, Web research, and more reading. I still take notes on 4×6 notecards just as I did when I was trying to index those Latin sermons. Once I process the “note”, I throw it away. I am lucky because once I read, write, and integrate the factoid into something I am writing — I remember the information. I don’t use digital calendars. I don’t use integrated to do lists. I just do what has been old fashioned information acquisition work.
The computer is wonderful for writing, Web research, and cooking up PowerPoint pablum. But the idea that using a tool that generates incorrect information strikes me as plain crazy.
The write up says:
Longji Cuo, an associate professor at the University of Colorado, in Boulder, teaches a course on AI and machine learning to help mechanical engineering students learn to use the technology to solve real-world engineering problems. Cuo encourages students to use AI as an agent to help with teamwork, projects, coding, and presentations in class. “My expectation on the quality of the work is much higher,” Cuo said, adding that students need to “demonstrate creativity on the level of a senior-level doctoral student or equivalent.”
Maybe. I am not convinced. Engineering issues are cascading across current and new systems. AI doesn’t seem to stem the tide. What about AI cyber security? Yeah, it’s working great. What about coding assistants? Yeah, super. I just uninstalled another Microsoft Windows 11 update. This one can kill my data storage devices. Copilot? Yeah, wonderful.
The write up concludes with this assertion from an “expert”:
one day, AI agents will be able to work with students on their personalized education needs. “Rather than having one teacher for 30 students, you’ll have one AI agent personalized to each student that will guide them along.”
Learning is hard work. The silliness of computer aid instruction, laptops, iPads, mobile phones, etc. makes one thing clear, learning is not easy. A human must focus, develop discipline, refine native talents, demonstrate motivate, curiosity, and an ability to process information into something more useful than remembering the TikTok icon’s design.
I don’t buy this. I am glad I am old.
Stephen E Arnold, August 25, 2025
Copilot, Can You Crash That Financial Analysis?
August 22, 2025
No AI. Just a dinobaby working the old-fashioned way.
The ever-insouciant online service The Verge published a story about Microsoft, smart software, and Excel. “Microsoft Excel Adds Copilot AI to Help Fill in Spreadsheet Cells” reports:
Microsoft Excel is testing a new AI-powered function that can automatically fill cells in your spreadsheets, which is similar to the feature that Google Sheets rolled out in June.
Okay, quite specific intentionality: Fill in cells. And a dash of me-too. I like it.
However, the key statement in my opinion is:
The COPILOT function comes with a couple of limitations, as it can’t access information outside your spreadsheet, and you can only use it to calculate 100 functions every 10 minutes. Microsoft also warns against using the AI function for numerical calculations or in “high-stakes scenarios” with legal, regulatory, and compliance implications, as COPILOT “can give incorrect responses.”
I don’t want to make a big deal out of this passage, but I will do it anyway. First, Microsoft makes clear that the outputs can be incorrect. Second, don’t use it too much because I assume one will have to pay to use a system that “can give incorrect results.” In short, MSFT is throttling Excel’s Copilot. Doesn’t everyone want to explore numbers with an addled Copilot known to flub numbers in a jet aircraft at 0.8 Mach?
I want to quote from “It Took Many Years And Billions Of Dollars, But Microsoft Finally Invented A Calculator That Is Wrong Sometimes”:
Think of it. Forty-five hundred years ago, if you were a Sumerian scribe, while your calculations on the world’s first abacus might have been laborious, you could be assured they’d be correct. Four hundred years ago, if you were palling around with William Oughtred, his new slide rule may have been a bit intimidating at first, but you could know its output was correct. In the 1980s, you could have bought the cheapest, shittiest Casio-knockoff calculator you could find, and used it exclusively, for every day of the rest of your life, and never once would it give anything but a correct answer. You could use it today! But now we have Microsoft apparently determining that “unpredictability” was something that some number of its customers wanted in their calculators.
I know that I sure do. I want to use a tool that is likely to convert “high-stakes scenarios” into an embarrassing failure. I mean who does not want this type of digital Copilot?
Why do I find this Excel with Copilot software interesting?
- It illustrates that accuracy has given way to close enough for horseshoes. Impressive for a company that can issue an update that could kill one’s storage devices.
- Microsoft no longer dances around hallucinations. The company just says, “The outputs can be wrong.” But I wonder, “Does Microsoft really mean it?” What about Red Bull-fueled MBAs handling one’s retirement accounts? Yeah, those people will be really careful.
- The article does not come and and say, “Looks like the AI rocket ship is losing altitude.”
- I cannot imagine sitting in a meeting and observing the rationalizations offered to justify releasing a product known to make NUMERICAL errors.
Net net: We are learning about the quality of [a] managerial processes at Microsoft, [b] the judgment of employees, and [c] the sheer craziness that an attorney said, “Sure, release the product just include an upfront statement that it will make mistakes.” Nothing builds trust more than a company anchored in customer-centric values.
Stephen E Arnold, August 22, 2025
-
- Subscribe to Beyond Search
Feature archive
News archive
-
