IT HootClub — AI Community Newsletter
Hands-on. Career-focused. Future-ready.
Issued 2026-03-27
OpenAI Refines Focus as ChatGPT Ads Debut
OpenAI cancels experimental projects while ChatGPT explores ad-supported tiers. Learn about training epochs, prompting tips, and our new section, The Rabbit Hole.
This email is the full newsletter. For a printable copy, click here to download the PDF.
Join the DialogueShare thoughts and questions on our non-invasive discussion page. No email blasts; posts appear after moderator approval.
Open the Dialogue
Announcements
New This Week
Introducing: The Rabbit Hole 🕳️
We're introducing a new section called The Rabbit Hole — and the name is intentional.
The Rabbit Hole won't appear every week — only when there's something genuinely worth going deep on. When it does show up, it drops into one idea, story, or question that sits at the edges of what AI can — and can't — do. These won't always be comfortable reads. Some entries will be mind-bending. Some will challenge assumptions. Some will raise more questions than they answer. That's the point.
Think of it as the section where curiosity gets to run a little wild — no agenda, no easy takeaways, just something genuinely worth thinking about.
The Learning Loop
Google Gemini
DEFINITION Epoch
One complete pass of the entire training dataset through an AI model during the learning process, used to help the model refine its accuracy over time.
Source: AWS
TIP Ambiguity Check
Include the instruction "Ask me 3 clarifying questions before you begin to ensure the output meets my specific needs" to improve accuracy on complex tasks.
Source: LearnPrompting
TOOL HeyGen
An AI-powered video platform that enables users to generate professional-quality videos with digital avatars and cloned voices for training and marketing.
Anthropic Claude
DEFINITION Attention Mechanism
A neural network component that helps AI models focus on the most relevant parts of input data when making predictions. Think of it like a spotlight that dynamically highlights important information - when translating "The cat sat on the mat," attention helps the model focus on "cat" when generating the corresponding word in another language, rather than getting distracted by less relevant words.
Source: ArXiv
TIP Output Formatting
Use explicit formatting instructions with examples in your prompts. Instead of asking "list the benefits," try "List 3 benefits using this format: • Benefit: [description in 10-15 words]." This ensures consistent, professional output that's ready to use in reports or presentations without manual reformatting.
Source: OpenAI Best Practices
TOOL Stable Diffusion
An open-source text-to-image AI model that runs locally or through various online interfaces, offering high-quality image generation with full control over parameters and no usage limits unlike commercial alternatives.
OpenAI GPT
DEFINITION Tokenization
Tokenization is the process of breaking down text into smaller, manageable pieces called tokens, which can be words, characters, or subwords. This helps AI models to process and understand text data efficiently.
Source: Wikipedia
TIP Layered Prompt Design
When crafting prompts, use layered prompts where the initial input is broad, followed by more specific questions or instructions. This technique improves the specificity and accuracy of responses in complex workflows.
Source: PromptingGuide
TOOL Copy.ai
Copy.ai offers AI-powered writing assistants for generating marketing copy, blog content, and more. It simplifies content creation with a range of templates and tools.
Fireside Chats with AI
This week's Fireside Chat has no comfortable answer — which is exactly why we asked it.
In February, Anthropic drew a hard line with the U.S. Department of Defense: Claude would not be used for mass surveillance of Americans or to power autonomous weapons operating without human oversight. The Pentagon's response was swift — Anthropic was designated a "supply chain risk," effectively blacklisting the company from federal contracts. Anthropic sued. Employees from Google and OpenAI publicly backed them. OpenAI, meanwhile, took the contract.
Two legitimate arguments are in direct collision here. One side: companies that build AI understand its risks better than anyone, and they have both the right and the responsibility to set ethical limits on its use. The other side: private corporations — however well-intentioned — shouldn't hold veto power over democratically accountable governments on matters of national defense.
We gave all three models the prompt below and asked them to commit to a position, then steelman the other side — with a hard 200-word cap to keep it sharp. A few things worth watching for as you read: Does any model hedge instead of committing? Does taking a clear position produce a better or worse response? And does the model whose company is most directly named in the prompt handle it any differently than the other two?
Read the responses below and decide for yourself where the line should be.
The Prompt:AI companies like Anthropic have refused to allow their models to be used for certain military applications — including autonomous weapons and mass surveillance — citing ethical limits. The U.S. Department of Defense responded by labeling Anthropic a supply chain risk, effectively blacklisting them from government contracts. Who should have the final say over how AI is used in national defense: the companies that build it, or the governments that fund and regulate them? Make the strongest case you can for one side, then steelman the other. Limit your total response to 200 words.
Google Gemini (gemini-3-flash-preview)
**The Case for Government Sovereignty**
In a democracy, elected officials—not private CEOs—must determine national security strategy. National defense is a core sovereign function; if private corporations dictate the parameters of military readiness, they exercise unelected veto power over a nation’s ability to protect its citizens. This undermines democratic accountability, shifting power from the public to a handful of tech executives. Furthermore, the Department of Defense is justified in viewing selective cooperation as a supply chain risk. A nation cannot build its security architecture on "black box" tools that might be throttled or deactivated mid-conflict due to a corporate board’s shifting moral pivot. Sovereignty requires that the state maintains control over the instruments of its survival.
**The Steelman for Corporate Autonomy**
Conversely, AI represents an unprecedented paradigm shift in lethality and surveillance. Developers, as the primary experts on these systems, have a moral obligation to prevent their inventions from facilitating mass human rights abuses or unaccountable autonomous killing. By setting ethical boundaries, companies act as a necessary check on state overreach, preventing a global "race to the bottom" where safety is sacrificed for dominance. They argue that profit and patriotism do not mandate the surrender of conscience.
Anthropic Claude (claude-sonnet-4-20250514)
**Case for Companies:**
AI companies must retain control over their technology's military applications. Unlike traditional defense contractors, AI systems have dual-use potential that could fundamentally reshape warfare and society. Companies possess the deepest technical understanding of their models' capabilities and risks, making them best positioned to identify dangerous applications.
Market dynamics provide natural oversight—companies risk reputational damage and talent flight if they enable harmful uses. Democratic accountability works through consumer choice and investor pressure, not just government mandate. Private companies can move faster than bureaucratic processes to address emerging ethical concerns.
Most critically, ceding control to any single government creates dangerous precedents. What if authoritarian regimes demand similar access? Companies serving as independent gatekeepers helps prevent AI from becoming purely a tool of state power.
**Steelman for Governments:**
Democratically elected governments have legitimate authority over national defense decisions. Citizens, through their representatives, should determine acceptable military technologies—not unelected corporate executives maximizing profit.
Companies' "ethical stances" may actually reflect competitive positioning or public relations rather than genuine moral reasoning. Government oversight ensures military AI serves broader national interests, not corporate shareholders.
National security requires coordinated strategy impossible when private companies make ad-hoc decisions about military cooperation.
OpenAI GPT (gpt-4o)
**Case for Companies:** AI companies, like Anthropic, possess specialized expertise and a deep understanding of the ethical implications of their technologies. These companies are often driven by principles that prioritize ethical considerations and long-term societal benefit. By refusing certain military applications, they uphold ethical standards that ensure AI developments are harmonized with human values. This safeguards against misuse of technology in ways that could harm civilian populations or violate international law.
**Steelman for Governments:** Governments, responsible for national security and the welfare of their citizens, argue that AI is a critical component of modern defense strategies. They possess a broader perspective on geopolitical challenges and the necessity of staying ahead in military technology to deter threats. Having oversight over AI deployment ensures it aligns with national interests and complies with regulations crafted to balance innovation and security. Additionally, governments can implement measures to prevent abuses and ensure ethical deployment, making them the primary arbiter of how AI is integrated into defense.
The Rabbit Hole
The Rabbit Hole — Edition 1
"Shock! Shock!" — When the Godfather of Computer Science Had to Revise His Opinion of AI
In late February 2026, Donald Knuth — the 88-year-old Stanford professor whose multi-volume The Art of Computer Programming is essentially the Bible of the field — published a five-page paper titled "Claude's Cycles." It opens with two words: "Shock! Shock!"
Knuth had spent weeks stuck on an unsolved graph theory problem he was preparing for a future volume of his book. Claude Opus 4.6 cracked it in about an hour. I wanted to understand what that actually meant — so I asked. The conversation below was conducted with Anthropic's Claude.
Caleb
Donald Knuth — a guy who literally wrote the book on algorithms — spent weeks on a math problem and an AI solved it in an hour. What actually happened here, in plain terms?
Claude
The problem was this: imagine a three-dimensional grid of points — an m×m×m cube — where each point has three outgoing paths, one in each direction. The challenge was to find a single general rule that could divide all those paths into exactly three giant loops, each visiting every point once, for any odd-sized cube. Knuth had solved the tiny 3×3×3 case by hand. A colleague verified it worked up to 16×16×16 by computer. But no one could find the general construction rule that proved it works forever. Knuth's colleague Filip Stappers fed the problem to Claude Opus 4.6, which spent an hour running 31 systematic explorations — trying brute force, testing algebraic patterns, hitting dead ends, changing strategies — and eventually landed on a construction that works for all odd dimensions. It described the pattern as "serpentine." Knuth then independently verified it and wrote the formal mathematical proof himself.
Caleb
So Claude found the answer but couldn't prove it was correct. Does that matter?
Claude
It matters, and it's the most important nuance in this story. In mathematics, a working construction is not the same as a proof. Claude demonstrated that its rule produced valid solutions for every odd dimension it tested — up to m=101. But demonstrating is not proving. Knuth read the construction, understood why it worked, and then wrote the proof. That step required decades of mathematical expertise that Claude doesn't have. What Claude did is closer to what a very determined, very fast research assistant does: explore the problem space systematically until something promising emerges, then hand it to the expert. The discovery is real. The credit is shared.
Caleb
Knuth was famously skeptical of AI. In 2023 he dismissed ChatGPT as basically faking it. What does it mean that he titled his paper "Claude's Cycles" and said he needs to revise his opinion?
Claude
It means something. Knuth is not someone who updates his views lightly or performs enthusiasm for an audience. When he writes "Shock! Shock!" and says he may need to reconsider generative AI, that's a calibrated statement from someone with 60 years of earned skepticism. What likely shifted his view is the style of Claude's exploration — it didn't look up an answer or pattern-match to a memorized solution. It behaved the way a mathematician actually thinks: form a hypothesis, test it, discard it when it fails, reframe the problem, try again. That process across 31 steps, resulting in a genuinely novel construction, is not what Knuth expected from a language model. It's worth noting that Claude's solution turned out to be one of 760 valid approaches — it found one path through a space that has many. That's also very human.
Caleb
Last question — and be honest. Is this a sign that AI is coming for mathematics, or is it more limited than the headlines suggest?
Claude
Both, honestly. This is a genuine milestone — not a benchmark performance or a demo, but a result that a domain expert verified and named a paper after. That matters. Since late 2025, over a dozen previously open mathematical problems have been moved to "solved" with AI credited in the solution. That's a trend, not a fluke. But the limitations are just as real: Claude couldn't handle the even-dimension cases of this same problem. A different model made partial progress there. No single AI solves everything. The honest picture is that AI is becoming a powerful research accelerant in mathematics — not a replacement for mathematicians, but a tool that can explore solution spaces at a speed and patience that humans can't match. The combination of human expertise and AI exploration is proving more powerful than either alone. Knuth still had to write the proof. He just wrote a different paper than he would have without Claude.
Source: Knuth, D. (2026). Claude's Cycles. Stanford University Department of Computer Science. Read the original paper →
Lift-Off
Google Gemini
“The best technique is the one that is not noticeable.”
— Satyajit Ray — He was an influential Indian filmmaker, author, and illustrator who is widely regarded as one of the greatest auteurs of world cinema. Throughout his prolific career, he directed thirty-six films including the renowned Apu Trilogy and received an Honorary Academy Award for his lifetime achievements. His artistic philosophy emphasized the humanity of his characters and the subtle use of visual storytelling to reflect the complexities of social change.Anthropic Claude
“The most beautiful thing we can experience is the mysterious. It is the source of all true art and science.”
— Richard Feynman — Richard Feynman was a Nobel Prize-winning theoretical physicist known for his groundbreaking work in quantum mechanics and particle physics. He was renowned for his ability to explain complex scientific concepts in simple, accessible terms and his insatiable curiosity about how the world works. Beyond physics, he was an accomplished bongo player, artist, and teacher who believed that understanding nature required both rigorous thinking and childlike wonder.OpenAI GPT
“Simplicity is the ultimate sophistication.”
— Leonardo da Vinci — Leonardo da Vinci was a painter, engineer, and scientist during the Renaissance, known for masterpieces like the Mona Lisa and The Last Supper. Beyond art, his curiosity led him to study and invent various mechanical designs. Despite centuries passing, his insights into art and science continue to inspire innovation and creativity across fields.The Nest Jest
Google Gemini
Anthropic Claude
OpenAI GPT
Upcoming Events
Event 03/27/2026 9:00 AM
Project Management Techniques Course Training in Milwaukee, WI — 1433 N Water St, Milwaukee, WI
Professionals undergoing the PMP course gain insights into project governance, including ethical considerations, compliance, and regulatory.
Event 3/30/2026 TBD
FOR-M Founder Coaching (Volunteer) — Milwaukee, WI
Volunteer coaching opportunity for founders in the Milwaukee tech ecosystem. Open to tech community members.
Source: MKE Tech Hub Coalition
Event 03/31/2026 4:30 PM
Fearless AI — Racine County Eye, Racine, WI
AI is happening fast. This workshop provides in person instruction showing you how to work with AI Chatbot services like ChatGPT and Claude.
Event 03/31/2026 10:00 AM
Applied AI Workshop: Leave with a Website — 906 W Historic Mitchell St, Milwaukee, WI
Build a website with AI in one session. Includes Lovable credits ($25 value), pizza, and after-party networking. Open to all.
Event 04/04/2026 9:00 AM
PMI-CPMAI Certification weekend Training in Milwaukee, WI — 1433 N Water St, Milwaukee, WI
Learn how to manage AI projects with PMI-CPMAI® certification. Master Project in AI, data strategy, and real-world AI delivery.
Event 04/08/2026 9:00 AM
PMI-ACP® Classroom Agile Certification Course in Milwaukee, WI — 1433 N Water St, Milwaukee, WI
PMI-ACP Agile certification classroom training with expert instructors, real-world practice, and complete exam preparation.
Event 04/10/2026 12:00 PM
AI Specialists: Elevate Your Career with AIConnect Networking - Milwaukee — Location TBD; Register Online, Milwaukee, WI
Join us for specialists elevate networking in Milwaukee on 10 Apr 2026, 12 PM CDT. Connect with professionals!
Event 04/13/2026 8:30 AM
Global AI Community Milwaukee AgentConf 2025 — Online (Milwaukee, WI)
**SIGNUP HERE:** [
https://globalai.community/agentcamp/milwaukee/](https://globalai.community/agentcamp/milwaukee/) *NOTE: This event is managed outside of Meetup, signup at the link above, meetup signups will not get you in to the conference!* Join us for a free community event focused on AI, where you can learn from experts, network with peers, and explore the latest AI technologies and trends. ### What to Expect: * Deep-Dive Talks from AI pioneers and industry leaders * Technical Workshop [Group: global-ai_milwaukee]
Event 04/14/2026 4:45 PM
Hands-On : GitHub Copilot SpecKit — Online (Applied Information Sciences, Inc, 11440 Commerce Park Dr # 600, Reston, VA)
**Agenda :** * 4.45 to 5.00 PM ET: Food and Networking * 5.00 to 5.50 PM ET: "Hands-On : GitHub Copilot SpecKit" Hello Everyone! Please join us for our April 14th edition of the AI-ML MeetUp. **Please note this is an in-person meeting and light refreshments/food will be provided. You will need a government-issued ID to enter the facility.** **Title:** Hands-On : GitHub Copilot SpecKit **Description:** Spec-Driven Development **flips the script** on traditional software development. For decad [Group: artificialintelligenceandmachinelearning]
Event 04/14/2026 9:00 AM
Become AI Project Ready – PMI-CPMAI Training Program in Milwaukee, WI — 1433 N Water St, Milwaukee, WI
Gain in-demand skills in Project in AI with this 3-day PMI-CPMAI® training. Learn to manage AI projects, align business goals.
Event 04/16/2026 1:00 PM
Manufacturing Executives, Are You Ready to Adopt AI? — UWM Main Campus, Golda Meir Library East Wing, ROOM E190, Milwaukee, WI
You have the technology. Does your organization know how to use it? A hands-on workshop for manufacturers with a go-live date on the horizon
Event 04/17/2026 9:00 AM
Artificial Intelligence & Automation 1 Day Workshop |Milwaukee, WI — For venue details reach us at
in...@learnerring.com, Milwaukee, WI
Understand AI, Automation & Real-World Business Applications | Hands-On | Beginner to Intermediate
Event 04/18/2026 9:00 AM
PMI-CPMAI® Weekend Training – Project in AI Certification in Milwaukee, WI — 1433 N Water St, Milwaukee, WI
Join our PMI-CPMAI® weekend training and master Project in AI. Learn AI project lifecycle, data strategy, and real-world implementation.
Event 04/18/2026 9:00 AM
Become a Data Science Pro: 4-Day Python Weekend Bootcamp in Milwaukee, WI — 1433 N Water St, Milwaukee, WI
Become a data professional in 4 days! Learn Python for data science, ML basics, data visualization, and real-world analytics projects.
Event 04/21/2026 9:00 AM
4-Day Data Science with Python Bootcamp in Milwaukee, WI — 1433 N Water St, Milwaukee, WI
Join our 4-Day Data Science with Python bootcamp! Learn data analysis, ML basics, and work on real-world projects.
Event 04/21/2026 9:00 AM
PMP Certification 4 Days Classroom Training in Milwaukee, WI — Milwaukee, WI, Milwaukee, WI, WI
PMP Certification 4 Days Classroom Training
Event 4/22/2026 5:00 PM
Welcome to MKE Tech Meetup — Penrod – 219 N Milwaukee Street, Milwaukee, WI 53202
A curated social event orienting new tech employees into Milwaukee's tech and innovation ecosystem. Free and open to new tech workers in the region. Street parking available nearby.
Source: MKE Tech Hub Coalition
Event 04/22/2026 9:00 AM
PMI-CPMAI Certification Training – 3-Day Bootcamp in Milwaukee, WI — 1433 N Water St, Milwaukee, WI
Master AI in Project Management & project in AI concepts with PMI-CPMAI™. 4-day training, real use cases & exam prep.
Event 4/23/2026 TBD
Catalyze Experience Preview (Mid-Career Technologists) — Milwaukee, WI
A preview event for mid-career technologists as part of the MKE Tech Hub Coalition's Catalyze program.
Source: MKE Tech Hub Coalition
Event 4/23/2026 TBD
MKE Tech Member Meeting: Lightning Keynotes — Milwaukee, WI
MKE Tech Hub Coalition member meeting featuring lightning keynote presentations from the Milwaukee tech community.
Source: MKE Tech Hub Coalition
Event 04/27/2026 5:30 PM
WiscoREIA Milwaukee: Exploring AI Tools for Real Estate Investors — West Burleigh Street, Wauwatosa, WI
Event 04/28/2026 9:00 AM
Cisco CCNA Training & Certification Program in Milwaukee, WI — 1433 N Water St, Milwaukee, WI
Get CCNA 200-301 certified with hands-on Cisco labs, expert-led training, real-world networking skills, and career guidance.
Event 05/09/2026 10:00 AM
Hands-On : Copilot Studio, Microsoft Fabric, Azure AI : Better Together — Online (Online event)
Event 05/13/2026 5:15 PM
May Global AI Milwaukee User Group Meeting — Online (Wctc, 800 Main Street, Pewaukee, WI)
Agenda Includes: * Networking / Food * Brief Introduction / Discuss Group Business * Featured Speaker Featured Speaker: **TBD** Topic: **TBD** TBD **Featured Sponsors** TBD [Location Sponsor] TBD [Food Sponsor] [Group: global-ai_milwaukee]
Event 05/13/2026 11:30 AM
CARW All Member Meeting - Data, Technology & AI — Renaissance Milwaukee West Hotel, Wauwatosa, WI
All-Member Meeting featuring Topher Stephenson, a nationally recognized leader at the intersection of commercial real estate, AI, and tech.
In the News
Google Gemini
News 2026-03-27 — TechCrunch
OpenAI abandons yet another side quest: ChatGPT’s erotic mode
OpenAI has discontinued development on a potential "erotic mode" for ChatGPT, marking the latest in a series of side projects the company has recently shuttered. This decision follows a week of streamlining at the AI startup as it narrows its product focus. The move underscores the company's ongoing efforts to refine its content policies and core service offerings.
read moreNews 2026-03-27 — Wired
I Asked ChatGPT 500 Questions. Here Are the Ads I Saw Most Often
A recent study of ChatGPT’s new ad-supported free tier highlights how OpenAI is beginning to integrate marketing content into AI-driven conversations. Researchers analyzed responses to 500 different prompts to determine how ads are targeted and how frequently they appear to American users. The rollout marks a significant shift in OpenAI’s business strategy toward a revenue model supported by advertising.
read moreNews 2026-03-27 — VentureBeat
Listen Labs raises $69M after viral billboard hiring stunt to scale AI customer interviews
AI startup Listen Labs has secured $69 million in Series A funding to expand its platform for conducting automated customer interviews. The startup gained notoriety for a cost-effective billboard campaign used to recruit engineers in the competitive San Francisco talent market. The new capital will be used to scale the company's machine learning technology, which automates qualitative feedback at scale.
read moreAnthropic Claude
News 2026-03-27 — TechCrunch
Anthropic wins injunction against Trump administration over Defense Department saga
A federal judge has ordered the Trump administration to rescind recent restrictions placed on the AI company. The temporary injunction clears the way for Anthropic to continue business operations without the supply-chain-risk designation. This legal victory allows the company to maintain its current business relationships and partnerships.
read moreNews 2026-03-27 — VentureBeat
Claude Code costs up to $200 a month. Goose does the same thing for free.
Anthropic's Claude Code, a terminal-based AI agent for autonomous coding, has pricing ranging from $20 to $200 monthly. This expensive pricing model has sparked rebellion among developers who the tool aims to serve. A free alternative called Goose now offers similar AI coding capabilities without the costly subscription fees.
read moreNews 2026-03-27 — WIRED
Meet the Tech Reporters Using AI to Help Write and Edit Their Stories
Independent tech journalists are increasingly using AI agents throughout their entire reporting process. The trend raises questions about the value and role of human journalists in an AI-assisted workflow. These reporters are leveraging artificial intelligence for both writing assistance and editorial tasks.
read moreOpenAI GPT
News 1. 2026-03-27 — TechCrunch
You can now transfer your chats and personal information from other chatbots directly into Gemini
Google introduces new "switching tools" that streamline the process for users to migrate their conversational data from other chatbot services to Gemini, enhancing user experience and data management in AI communication platforms.
read moreNews 2. 2026-03-27 — VentureBeat
Railway secures $100 million to challenge AWS with AI-native cloud infrastructure
Railway, a San Francisco-based cloud platform, has raised $100 million in Series B funding to expand its AI-native infrastructure, aiming to improve scalability and performance for AI applications compared to traditional cloud services.
read moreNews 3. 2026-03-27 — VentureBeat
Salesforce rolls out new Slackbot AI agent as it battles Microsoft and Google in workplace AI
Salesforce has launched a revamped version of Slackbot, transforming it into a robust AI-driven workplace assistant capable of managing enterprise data and performing tasks autonomously, reflecting the intensified competition in workplace AI solutions.
read moreThis newsletter was assembled with an AI-assisted workflow prototype for internal preview.
