Anthropic Discovers a Moral Code in Its Smart Software
April 30, 2025
No AI. This old dinobaby just plods along, delighted he is old and this craziness will soon be left behind. What about you?
With the United Arab Emirates starting to use smart software to make its laws, the idea that artificial intelligence has a sense of morality is reassuring. Who would want a person judged guilty by a machine to face incarceration, a fine, or — gulp! — worse.
“Anthropic Just Analyzed 700,000 Claude Conversations — And Found Its AI Has a Moral Code of Its Own” explains:
The [Anthropic] study examined 700,000 anonymized conversations, finding that Claude largely upholds the company’s “helpful, honest, harmless” framework while adapting its values to different contexts — from relationship advice to historical analysis. This represents one of the most ambitious attempts to empirically evaluate whether an AI system’s behavior in the wild matches its intended design.
Two philosophers watch as the smart software explains the meaning of “situational and hallucinatory ethics.” Thanks, OpenAI. I bet you are glad those former employees of yours quit. Imagine. Ethics and morality getting in the way of accelerationism.
Plus the company has “hope”, saying:
“Our hope is that this research encourages other AI labs to conduct similar research into their models’ values,” said Saffron Huang, a member of Anthropic’s Societal Impacts team who worked on the study, in an interview with VentureBeat. “Measuring an AI system’s values is core to alignment research and understanding if a model is actually aligned with its training.”
The study is definitely not part of the firm’s marketing campaign. The write up includes this quote from an Anthropic wizard:
The research arrives at a critical moment for Anthropic, which recently launched “Claude Max,” a premium $200 monthly subscription tier aimed at competing with OpenAI’s similar offering. The company has also expanded Claude’s capabilities to include Google Workspace integration and autonomous research functions, positioning it as “a true virtual collaborator” for enterprise users, according to recent announcements.
For $2,400 per year, a user of the smart software would not want to do something improper, immoral, unethical, or just plain bad. I know that humans have some difficulty defining these terms related to human behavior in simple terms. It is a step forward that software has the meanings and can apply them. And for $200 a month one wants good judgment.
Does Claude hallucinate? Is the Anthropic-run study objective? Are the data reproducible?
Hey, no, no, no. What do you expect in the dog-eat-dog world of smart software?
Here’s a statement from the write up that pushes aside my trivial questions:
The study found that Claude generally adheres to Anthropic’s prosocial aspirations, emphasizing values like “user enablement,” “epistemic humility,” and “patient wellbeing” across diverse interactions. However, researchers also discovered troubling instances where Claude expressed values contrary to its training.
Yes, pro-social. That’s a concept definitely relevant to certain prompts sent to Anthropic’s system.
Are the moral predilections consistent?
Of course not. The write up says:
Perhaps most fascinating was the discovery that Claude’s expressed values shift contextually, mirroring human behavior. When users sought relationship guidance, Claude emphasized “healthy boundaries” and “mutual respect.” For historical event analysis, “historical accuracy” took precedence.
Yes, inconsistency depending upon the prompt. Perfect.
Why does this occur? This statement reveals the depth and elegance of the Anthropic research into computer systems whose inner workings are tough for their developers to understand:
Anthropic’s values study builds on the company’s broader efforts to demystify large language models through what it calls “mechanistic interpretability” — essentially reverse-engineering AI systems to understand their inner workings. Last month, Anthropic researchers published groundbreaking work that used what they described as a “microscope” to track Claude’s decision-making processes. The technique revealed counterintuitive behaviors, including Claude planning ahead when composing poetry and using unconventional problem-solving approaches for basic math.
Several observations:
- Unlike Google which is just saying, “We are the leaders,” Anthropic wants to be the good guys, explaining how its smart software is sensitive to squishy human values
- The write up itself is a content marketing gem
- There is scant evidence that the description of the Anthropic “findings” are reliable.
Let’s slap this Anthropic software into an autonomous drone and let it loose. It will be the AI system able to make those subjective decisions. Light it up and launch.
Stephen E Arnold, April 30, 2025
Want Traffic from Google? Buy Ads, Lots of Ads
April 30, 2025
No AI, just the dinobaby himself.
When I was working, clients and prospects would ask me, “Do I need to advertise on Google to get traffic to my Web site?” I relayed the “facts” as I understood them at the time. My answer was, “You need to buy ads from Google.”
Most of the clients wrinkled their foreheads and asked, “Why?” My answer was then and still is, “Do you think Google does things for you for free?” Since I don’t do advertising, I don’t know how my information filtered from my contacts to the people who handled these organizations’ advertising budgets. I knew that with big indexes and lots of users, only a tiny fraction of the terms and Web sites get traffic. People don’t understand that their Web site is mostly invisible and was destined to stay that way unless [a] something extraordinary appeared on a Web page and drew eyeballs or [b] the organization had to spend thousands each month on Google ads.
I thought times might be changing since I retired. Nope, advertising matters. If the information in “Temu Pulls Its US Google Shopping Ads” is accurate, Google ads matter. The article reports:
Temu completely shut off Google Shopping ads in the U.S. on April 9, with its App Store ranking subsequently plummeting from a typical third or fourth position to 58th in just three days. The company’s impression share, which measures how often their ads appear compared to eligibility, dropped sharply before disappearing completely from advertiser auction data by April 12.
Buy ads, get traffic. That was true when I was running myself ragged trying to do work, and it is true today. I would suggest that this Temu example offers some insight into what happens if apps get pulled from the Google Play Store. Whatever downloads a developer had are likely to take a hit; that is, go from hero to zero in a snap.
The article wanders into political issues which are not part of my job description. I think it is important to think in terms of findability. One can pray that one’s content is so darned compelling that people flock to a magnet site or a post. Hope springs eternal just like every baby is a genius. One can pay search engine optimization wizards to gin up traffic via white hat and black hat methods. One can just buy ads, and go with the pay-to-play method.
Am I okay with Google’s control of traffic? Sure. I don’t care if I get traffic. But others do and need traffic to stay in business. Therefore, the information about Temu is germane I think. Your baby is a genius. Believe that. Just don’t assume that traffic will automatically flow to that baby’s Web site even if you bought a domain name celebrating the birth. Just buy ads.
Stephen E Arnold, April 30, 2025
France And Germany Form Open Source Writing Collaboration
April 30, 2025
Open source software and AI algorithms are a match made in heaven. You can’t say the same thing about France and Germany when it comes to history, but the countries can put aside their differences (occasionally) to advance technology. The French and German governments came together to design Docs.
Docs is described as “Collaborative writing, simplified-collaborate and write in real time, without layout constraints.” I don’t know if the term “layout” refers to a writing software’s formatting or if it means limited to the constraints of writing software. It could mean either of things or something is lost in the literal translation. Ich habe keine Ahnung. Je ne c’est pas.
Docs is built on the Django Rest Framework and Nest.js. It also uses BlockNote.js and Yes (they also sponsor those text editors too). Docs can be self-hosted, has a business friendly license, and welcomes anyone to contribute to its growth either monetarily or via code). Here is what Docs offers as a writing partner:
“Docs offers an intuitive writing experience. Its minimalist interface favors content over layout, while offering the essentials: media import, offline mode and keyboard shortcuts for greater efficiency.”
So far that sounds très magnifique and ausgezeichnet! Docs also offers simple real-time collaboration. Users on a document can access the same document, see changes made live, and maintain control of the document for data security. Docs also has universal formats for exportation: OpenDocument, Word, and PDF.
A nifty feature unavailable with most writing software is the ability to organize documents into knowledge bases with subpages. This feature also comes with search and pinning capabilities.
This French and German writing collaboration sounds amazing! Break out the champagne and beer and enjoy some croissants and pretzels. This is one open source tool everyone needs!
Whitney Grace, April 30, 2025
Google Wins AI, According to Google AI
April 29, 2025
No AI. This old dinobaby just plods along, delighted he is old and this craziness will soon be left behind. What about you?
Wow, not even insecure pop stars explain how wonderful they are at every opportunity. But Google is not going to stop explaining that it is number one in smart software. Never mind the lawsuits. Never mind the Deepseek thing. Never mind Sam AI-Man. Never mind angry Googlers who think the company will destroy the world.
Just get the message, “We have won.”
I know this because I read the weird PR interview called “Demis Hassabis Is Preparing for AI’s Endgame,” which is part of the “news” about the Time 100 most wonderful and intelligence and influential and talented and prescient people in the Time world.
Let’s take a quick look at a few of the statements in the marketing story. Because I am a dinobaby, I will wrap up with a few observations designed to make clear the difference between old geezers like me and the youthful new breed of Time leaders.
Here’s the first passage I noted:
He believes AGI [Googler Hassabis] would be a technology that could not only solve existing problems, but also come up with entirely new explanations for the universe. A test for its existence might be whether a system could come up with general relativity with only the information Einstein had access to; or if it could not only solve a longstanding hypothesis in mathematics, but theorize an entirely new one. “I identify myself as a scientist first and foremost,” Hassabis says. “The whole reason I’m doing everything I’ve done in my life is in the pursuit of knowledge and trying to understand the world around us.”
First comment. Yep, I noticed the reference to Einstein. That’s reasonable intellectual territory for a Googler. I want to point out that the Google is in a bit of legal trouble because it did not play fair. But neither did Einstein. Instead of fighting evil in Europe, he lit out for the US of A. I mean a genius of the Einstein ilk is not going to risk one’s life. Just think. Google is a thinking outfit, but I would suggest that its brush with authorities is different from Einstein’s. But a scientist working at an outfit in trouble with authorities, no big deal, right? AI is a way to understand the world around us. Breaking the law? What?
The second snippet is this one:
When DeepMind was acquired by Google in 2014, Hassabis insisted on a contractual firewall: a clause explicitly prohibiting his technology from being used for military applications. It was a red line that reflected his vision of AI as humanity’s scientific savior, not a weapon of war.
Well, that red line was made of erasable market red. It has disappeared. And where is the Nobel prize winner? Still at the Google, that’s the outfit that is in trouble with the law and reasonably good at discarding notions that don’t fit with its goal of generating big revenue from ads and assorted other ventures like self driving taxi cabs. Noble indeed.
Okay, here’s the third comment:
That work [dumping humans for smart software], he says, is not intended to hasten labor disruptions, but instead is about building the necessary scaffolding for the type of AI that he hopes will one day make its own scientific discoveries. Still, as research into these AI “agents” progresses, Hassabis says, expect them to be able to carry out increasingly more complex tasks independently. (An AI agent that can meaningfully automate the job of further AI research, he predicts, is “a few years away.”)
I think that Google will just say, “Yo, dudes, smart software is efficient. Those who lose their jobs can re-skill like the humanoids we are allowing to find their future elsewhere.
Several observations:
- I think that the Time people are trying to balance their fear of smart software replacing outfits like Time with the excitement of watching smart software create a new way experiencing making a life. I don’t think the Timers achieved their goal.
- The message that Google thinks, cares, and has lofty goals just doesn’t ring true. Google is in trouble with the law for a reason. It was smart enough to make money, but it was not smart enough to avoid honking off regulators in some jurisdictions. I can’t reconcile illegal behavior with baloney about the good of mankind.
- Google wants to be seen as the big dog of AI. The problem is that saying something is different from the reality of trials, loss of trust among some customer sectors, floundering for a coherent message about smart software, and the baloney that the quantumly supreme Google convinces people to propagate.
Okay, you may love the Time write up. I am amused, and I think some of the lingo will find its way into the Sundar & Prabhakar Comedy Show. Did you hear the one about Google’s AI not being used for weapons?
Stephen E Arnold, April 29, 2025
Apple and Meta: Virtual Automatic Teller Machines for the EU
April 29, 2025
No AI, just a dinobaby watching the world respond to the tech bros.
I spotted this story in USA Today. You remember that newspaper, of course. The story “Apple Fined $570 Million and Meta $228 Million for Breaching European Union Law” reports:
Apple was fined 500 million euros ($570 million) on Wednesday and Meta 200 million euros, as European Union antitrust regulators handed out the first sanctions under landmark legislation aimed at curbing the power of Big Tech.
I have observed that to many regulators the brands Apple and Meta (Facebook) are converted to the sound of ka-ching. For those who don’t recognize the onomatopoeia for an old-fashioned cash register ringing up a sale. The modern metaphor might be an automatic teller machine emitting beeps and honks. That works. Punch the Apple and Meta logos and bonk, beep, out comes millions of euros. Bonk, beep.
The law which allows the behavior of what some Europeans view as “tech bros” to be converted first to a legal process and then to cash is the Digital Markets Act. The idea is that certain technology centric outfits based in the US operate without much regard for the rules, regulations, and laws of actual nation-states and their governing entities. I mean who pays attention to what the European Union says? Certainly not a geek à la sauce californienne.
The companies are likely to interpret these fines as some sort of deus ex machina, delivered by a third-rate vengeful god in a TikTok-type of video. Perhaps? But the legal process identified some actions by the fined American companies as illegal. Examples range from preventing an Apple store user from certain behaviors to Meta’s reluctance to conform to some privacy requirements. I am certainly not a lawyer, nor am I involved with either of the American companies. However, I can make several observations from my dinobaby point of view, of course:
- The ka-ching / bonk beep incentive is strong. Money talks in the US and elsewhere. It is not surprising that the fines are becoming larger with each go-round. How does one stop the cost creep? One thought is to change the behavior of the companies. Sorry, EU, that is not going to happen.
- The interpretation of the penalty as a reaction against America is definitely a factor. For those who have not lived and worked in other countries, the anti-American sentiment is not understood. I learned when people painted slurs on the walls of our home in Campinas, Brazil. I was about 13, and the anger extended beyond black paint on our pristine white, eight-foot high walls with glass embedded at the top of them. Inviting, right?
- The perception that a company is more powerful than a mere government entity has been growing as the concentration of eyeballs, money, and talented people has increased at certain firms. Once the regulators have worked through the others in this category, attention will turn to the second tier of companies. I won’t identify any entities but the increased scrutiny of Cloudflare by French authorities is a glimpse of what might be coming down the information highway.
Net net: Ka-ching, ka-ching, and ka-ching. Beep, bong, beep, bong.
Stephen E Arnold, April 29, 2025
China, Self-Amusement, and AI
April 29, 2025
China pokes fun at the United States whenever it can. Why? The Middle Kingdom wants to prove its superiority over the US. China is does have many technological advances over its western neighbor and now the country made another great leap forward with AI says Business Insider: “China’s Baidu Releases Ernie X1, A New AI Reasoning Model.”
Baidu is China’s equivalent of Google and the it released two new AI models. The first is Ernie X1 that is described as a reasoning model that delivers on par with Deepseek R1 at half the price. It also released a multimodal foundation model called Ernie 4.5 that could potentially outperform GPT-4.5 and costs only a fraction of the price. Baidu is also developing the Ernie Bot, a free chatbot.
Baidu wants to offer the world cheap AI:
“Baidu’s new releases come as Silicon Valley reckons with the cost of AI models, largely spurred by the latest drops from Deepseek, a Chinese startup launched by hedge fund High Flyer.
In December, Deepseek released a large language model called V3, and in January, it unveiled a reasoning model called R1. The models are considered as good or better than equivalent models from OpenAI but priced “anywhere from 20-40x cheaper,” according to analysis from Bernstein Research.”
China is smart to develop inexpensive AI, but did the country have to make fun of Sesame Street? I mean Big Bird?
Whitney Grace, April 29, 2025
The Only-Google-Can-Do-It Information Campaign: Repeat It, and It Will Be “True.” Believe Now!
April 28, 2025
No AI. Just a dinobaby who gets revved up with buzzwords and baloney.
After more than two decades of stomping around the digital world, the Google faces some unpleasant consequences of what it hath wrought. There is the European Union’s ka-ching factor; that is, Google is a big automatic teller machine capable of spitting out oodles of cash after the lawyers run out of gas. The US legal process is looking more like the little engine that could. If it can, Google may lose control of some of its big-time components; for example, the Chrome browser. I think this was acquired by the Google from someone in Denmark years ago, but I am a bit fuzzy about this statement. But, hey, let’s roll with it. Google “owns” the browser market, and if the little engine that could gets to the top of the hill (not guaranteed by any means, of course) then another outfit might acquire it.
Among the players making noises about buying the Google browser is OpenAI. I find this interesting because [a] Sam AI-Man wants to build his version of Telegram and [b] he wants to make sure that lots of people use his firm’s / organization’s smart software. Buy Chrome and Sam has users and he can roll out a browser enabled version of the Telegram platform with his very own AI system within.
Google is not too keen on losing any of its “do good” systems. Chrome has been a useful vector for such helpful functions as data gathering, control of extensions, and having its own embedded Google search system everywhere the browser user goes. Who needs Firefox when Google has Chrome? Probably not Sam AI-Man or Yahoo or whoever eyes the browser.
“Only Google Can Run Chrome, Company’s Browser Chief Tells Judge” reveals to me how Google will argue against a decision forcing Google to sell its browser. That argument is, not surprisingly, is anchored within Google’s confidence in itself, its wizards, its money, and its infrastructure. The Los Angeles Times’ article says:
Google is the only company that can offer the level of features and functionality that its popular Chrome web browser has today, given its “interdependencies” on other parts of the Alphabet Inc. unit, the head of Chrome testified. “Chrome today represents 17 years of collaboration between the Chrome people” and the rest of Google, Parisa Tabriz, the browser’s general manager, said Friday as part of the Justice Department’s antitrust case in Washington federal court. “Trying to disentangle that is unprecedented.”
My interpretation of this comment is typical of a dinobaby. Google’s browser leader is saying, “Other companies are not Google; therefore, those companies are mentally, technically, and financially unable to do what Google does.” I understand. Googzilla is supreme in the way it is quantumly supreme in every advanced technology, including content marketing and public relations.
The write up adds:
James Mickens, a computer science expert for the Justice Department, said Google could easily transfer ownership of Chrome to another company without breaking its functionality. … “The divestiture of Chrome is feasible from a technical perspective,” said Mickens, a computer science professor at Harvard University. “It would be feasible to transfer ownership and not break too much.”
Professor Mickens has put himself in the category of non-Googley people who lack the intelligence to realize how incorrect his reasoning is. Too bad, professor, no Google consulting gig for you this year.
Plus, Google has a plan for its browser. The write up reports:
In internal documents, Google said it intends to develop Chrome into an “agentic browser,” which incorporates AI agents to automate tasks and perform actions such as filling out forms, conducting research or shopping. “We envision a future of multiple agents, where Chrome integrates deeply with Gemini as a primary agent and one we’ll prioritize and enable users to engage with multiple 3P agents on the web in both consumer and enterprise settings,” Tabriz wrote in a 2024 email.
How will this play out? I have learned that predicting the outcome of legal processes is a tough job. Stick to estimating the value of a TONcoin. That’s an easier task.
What does seem clear to me are three points:
- Google’s legal woes are not going away
- Google’s sense of its technology dominance is rising despite some signals that that perception may not align with what’s happening in AI and other technical fields
- Google’s argument that only it can do its browser may not fly in the midst of legal eagles.
I don’t think the “browser chief” will agree with this dinobaby. That’s okay. Trust me.
Stephen E Arnold, April 28, 2025
Japan Alleges Google Is a Monopoly Doing Monopolistic Things. What?
April 28, 2025
No AI, just the dinobaby himself.
The Google has been around a couple of decades or more. The company caught my attention, and I wrote three monographs for a now defunct publisher in a very damp part of England. These are now out of print, but their titles illustrate my perception of what I call affectionately Googzilla:
- The Google Legacy. I tried to explain why Google’s approach was going to define how future online companies built their technical plumbing. Yep, OpenAI in all its charm is a descendant of those smart lads, Messrs. Brin and Page.
- Google Version 2.0. I attempted to document the shift in technical focus from search relevance to a more invasive approach to using user data to generate revenue. The subtitle, I thought at the time, gave away the theme of the book: “The Calculating Predator.”
- Google: The Digital Gutenberg. I presented information about how Google’s “outputs” from search results to more sophisticated content structures like profiles of people, places, and things was preparing Google to reinvent publishing. I was correct because the new head of search (Prabhakar Version 2.0) is making little reports the big thing in search results. This will doom many small publications because Google just tells you what it wants you to know.
I wrote these monographs between 2002 and 2008. I must admit that my 300 page Enterprise Search Report sold more copies than my Google work. But I think my Google trilogy explained what Googzilla was doing. No one cared.
Now I learn “Japan orders Google to stop pushing smartphone makers to install its apps.”* Okay, a little slow on the trigger, but government officials in the land of the rising sun figured out that Google is doing what Google has been doing for decades.
Enlightenment arrives!
The article reports:
Japan has issued a cease-and-desist order telling Google to stop pressuring smartphone makers to preinstall its search services on Android phones. The Japan Fair Trade Commission said on Tuesday Google had unfairly hindered competition by asking for preferential treatment for its search and browser from smartphone makers in violation of the country’s anti-monopoly law. The antitrust watchdog said Google, as far back as July 2020, had asked at least six Android smartphone manufacturers to preinstall its apps when they signed the license for the American tech giant’s app store…
Google has been this rodeo before. At the end of a legal process, Google will apologize, write a check, and move on down the road.
The question for me is, “How many other countries will see Google as a check writing machine?”
Quite a few in my opinion. The only problem is that these actions have taken many years to move from the thrill of getting a Google mouse pad to actual governmental action. (The best Google freebie was its flashing LED pin. Mine corroded and no longer flashed. I dumped it.)
Note for the * — Links to Microsoft “news” stories often go dead. Suck it up and run a query for the title using Google, of course.
Stephen E Arnold, April 28, 2025
Banks and Security? Absolutely
April 28, 2025
The second-largest US bank has admitted it failed to recover documents lost to a recent data breach. The Daily Hodl reports, “Bank of America Discloses Data Breach After Customers’ Documents Disappear, Says Names, Addresses, Account Information and Social Security Numbers Affected.” Writer Mark Emem tells us:
“Bank of America says efforts to locate sensitive documents containing personal information on an undisclosed number of customers have failed. The North Carolina-based bank says it is unable to recover the documents, which were lost in transit and ‘resulted in the disclosure’ of personal information. [The bank’s notice states,] ‘According to our records, the information involved in this incident was related to your savings bonds and included your first and last name, address, phone number, Social Security number, and account number…We understand how upsetting this can be and sincerely apologize for this incident and any concerns or inconvenience it may cause. We are notifying you so we can work together to protect your personal and account information.’”
Banks are forthcoming and bad actors know there is money in them. It is no surprise Bank of America faces a challenge. The succinct write-up notes the bank’s pledge to notify affected customers of any suspicious activity on their accounts. It is also offering them a two-year membership to an identity theft protection service. We suggest any Bank of America customers go ahead and change their passwords as a precaution. Now. We will wait.
Cynthia Murrell, April 28, 2025
Geocoding Price Data
April 28, 2025
No AI, just a dinobaby watching the world respond to the tech bros.
Some data must be processed to add geocodes to the items. Typically the geocode is a latitude and longitude coordinate. Some specialized services add addition codes to facilitate height and depth measurements. However, a geocode is not just what are generally called “lat and long coordinates.” Here’s a selected list of some of the items which may be included in a for-fee service:
- An address
- A place identifier
- Boundaries like a neighborhood, county, etc.
- Time zone
- Points-of-interest data.
For organizations interested in adding geocodes to their information or data, pricing of commercial services becomes an important factor.
I want to suggest that you navigate to “Geocoding APIs Compared: Pricing, Free Tiers & Terms of Use.” This article was assembled in 2023. The fees presented are out of date. However, as you work through the article, you will gather useful information about vendors such as Google, MSFT Azure, and TomTom, among others.
One of the question-answering large language models can be tapped to provide pricing information that is more recent.
Stephen E Arnold, April 28, 2025
-
- Subscribe to Beyond Search
Feature archive
News archive
-
