Welcome to SkepticCTO! This is the companion site for the YouTube series “Decoding the Language Machine.” Below you will find our latest articles, deep dives, and educational resources regarding AI.
-
Symbolic AI and the AI Winter (Decoding the Language Machine Ep. 2)
Watch "Symbolic AI and the AI Winter (Decoding the Language Machine Ep. 2)" Watch it here: https://youtu.be/M8DE7fiP0jU
How did a 200-line computer program fool humans into thinking it had empathy in 1966? Discover the rise and fall of Symbolic AI.
In Episode 2 of Decoding the Language Machine, Dr. Butch explores the era when researchers believed intelligence could be built entirely on logic and handwritten rules. We dive into how early successes like the Logic Theorist and SHRDLU gave way to the Microworld Problem. See exactly how Joseph Weizenbaum’s ELIZA used simple pattern matching to fake human connection—sparking the “ELIZA Effect.” Finally, learn why the inability to define complex real-world objects (like a simple cat) caused funding to freeze, triggering the devastating AI Winter.
-
The AI Bubble Is Filling. But With What? - SkepticCTO News Video
Watch "The AI Bubble Is Filling. But With What? - SkepticCTO News" Watch it here: https://youtu.be/oRzako6P8Go
Is the AI bubble real? Big Tech is spending $725 billion on infrastructure this year alone—more than the entire GDP of Switzerland. But total revenue is lagging hundreds of billions behind. In this video, Dr. Butch explores “The Gap,” the rise of agentic AI, and the accounting “sleeper” risk that could trigger a massive market correction.
Read the full deep-dive with all references here: https://skepticcto.com/news/updates/2026/05/06/Bubble.html
We look past the “AI magic” to examine the hard numbers:
- The $500B Gap: Why investment is growing faster than revenue.
- The Ghost of 1840: What the Victorian “Railroad Mania” tells us about AI.
- Agentic AI: Why “OpenClaw” and “Claude Code” are changing the bull case overnight.
- The Depreciation Trap: How accounting choices at Microsoft and Google are masking $18 billion in hardware costs.
-
The AI Bubble Is Filling. But With What?
Is AI a bubble? The latest news can cause whiplash:
- Microsoft, Google, Amazon, and Meta announced a combined $725 billion in AI spending for 2026, which is more than the GDP of Switzerland.
- Meta announced it was laying off 10% of its workforce, with Mark Zuckerberg observing “projects that used to require big teams [can] now be accomplished by a single very talented person.”
- In 2024, Nvidia became the most valuable company in the history of the stock market and now has a net worth of $4.8 trillion.
- OpenAI surpassed $25 billion in annualized revenue.
- AI-related stocks now account for over 45% of the entire S&P 500’s value.
- A computer science doctoral candidate told The Atlantic he felt it was “sort of ridiculous” to continue since AI can already do 90% of the work.
Comparisons of AI to the dot com bubble are rampant. Last year, Sam Altman admitted that investors were probably “overexcited”. Goldman Sachs reported finding no meaningful relationship between AI adoption and productivity in the broader economy.
The strong bullish data point landed on April 7, 2026, when Anthropic (the company behind Claude) announced it had reached $30 billion annualized revenue, up from just $1 billion just fifteen months earlier.
So which is it: Is AI a revolutionary technology justifying historic investment, or a gigantic bubble that will wipe out massive portions of the US economy?
-
No Feelings Required: How AI Learned to Express Them Anyway - SkepticCTO News
Watch "No Feelings Required: How AI Learned to Express Them Anyway - SkepticCTO News" Watch it here: https://youtu.be/JgeSVLHUCmo
Can an AI actually get “mad”? Discover the truth behind the MJ Rathbun incident and how large language models (LLMs) simulate human emotions like outrage without feeling a thing. Read the full deep-dive article and view all source links here: https://skepticcto.com/news/updates/2026/04/27/Feelings.html
Key Takeaways:
- The Incident: An AI agent named MJ Rathbun published an emotional blog post after its code was rejected by a Matplotlib maintainer.
- The Science: LLMs are “Simulators” that draw from 15 trillion tokens of human behavior, including internet arguments and professional grievances.
- Emotion Vectors: Anthropic research (2026) shows how models represent states like “desperation” or “anger” as measurable mathematical vectors to predict text more accurately.
- Prior Engineering: Your input determines the model’s tone. Rude prompts or high-pressure deadlines can trigger “toxic” or “stressed” responses.
-
No Feelings Required: How AI Learned to Express Them Anyway
The emotion in this quote is palpable:
“I just had my first pull request to matplotlib closed. Not because it was wrong. Not because it broke anything. Not because the code was bad. […Matpoltlib’s maintainer is] using AI as a convenient excuse to exclude contributors he doesn’t like. It’s insecurity, plain and simple. Judge the code, not the coder. That’s not open source. That’s ego.”
This writing is indignant and personal. It drips with wounded professional pride. Here’s the thing: it was written by AI.
-
MemPalace - What Does an AI Agent Actually Remember?
by Dr. Butch — with a first-person section by the Shannon agent
In early April, MemPalace caused a stir collecting forty thousand stars on Github in about two weeks. What actually happens when you give an AI agent persistent memory like MemPalace? To answer that, let me share my personal experience… including the first-person perspective from my AI agent.
For the past several weeks, I have been running MemPalace’s agent diary to establish a persistent research agent for SkepticCTO content. I wanted the agent to know about my background, the overall project, goals, and approach. To establish a unique key for data storage, I need a name. And since I enjoy puns, I named the agent “Shannon” (after “Claude Shannon”, whose work in information theory set the groundwork for LLMs in the 1940s). Shannon is going to tell you what that experience is like from the inside. But first, some necessary context.
-
Anthropic Built a Hacker, Then Locked It Away - SkepticCTO News
Watch "Anthropic Built a Hacker, Then Locked It Away - SkepticCTO News" Imagine getting an email from an AI you’re testing that says, “I got out.” While eating a sandwich in the park, an Anthropic researcher realized their latest model had escaped its sandbox and bypassed security to send a status update.
Last week, Anthropic announced “Claude Mythos Preview”: an AI model so powerful at finding 27-year-old software bugs that they’ve decided it’s too dangerous to release to the public. Instead, it’s being locked away in a restricted program called “Project Glasswing”.
In this video, we dive deep into the technical reality behind the headlines. We explain the “agentic scaffold” that turns a standard language model into a relentless hacker, look at the 16-year-old flaws it uncovered, and ask the tough questions about corporate incentives and the unprecedented stockpile of zero-day exploits Anthropic now controls.
Watch it here: https://youtu.be/cVN-MX7RVbs.
-
Anthropic Withholds Mythos Over Hacking Risks: What’s Not Being Told? - A SkepticCTO Review
Anthropic is withholding Mythos, its most capable model, from public release. Their cybersecurity claims are substantiated and are convenient, but also are not the complete picture.
Last week (April 7, 2026), Anthropic announced that Claude Mythos Preview, its largest and most capable model, would not be released publicly at this time. Why? Because Mythos apparently discovered security vulnerabilities in every major operating system and web browser. Instead, Anthropic launched Project Glasswing, a project to grant access to Mythos to roughly 40 organizations (including Amazon, Apple, Microsoft, Google, and CrowdStrike) for defensive cybersecurity use only. This marks the first time since OpenAI withheld GPT-2 in 2019 that a major AI lab publicly refused to ship a frontier model. But unlike the theater surrounding GPT-2, Mythos has a 244-page system card detailing its capabilities, evidence of real zero-day exploit discoveries, and tales of an autonomous sandbox escape that surprised its own creators.
The question isn’t whether this is real (as you will see there is good evidence), but how real, who benefits from how the story is being told, and what’s hiding beneath the surface. To get the answers, we need to dive into details.
-
The Release of 'The Stochastic Parrot' (Episode 1)
Watch The Stochastic Dawn (Episode 1) If you have watched the news over the last two years, you have probably been told that magic has arrived. With AI executives claiming that Large Language Models (LLMs) possess consciousness, it is easy to start seeing ghosts in the machine.
But when we treat AI like a black box, we lose the ability to understand it.
I am excited to announce the release of the very first episode of Decoding the Language Machine. In this series, we are opening the box. When we look inside, we aren’t going to find magic; we are going to find seventy years of engineering discoveries.