Topic: Machine Learning

Fox Business host Dagan McDowell described an interaction with ChatGPT where she argued with the bot over its refusal to recognize Donald Trump's current term as a "second term." The segment illustrated the "Eliza Effect," where humans attribute human-like intelligence and intent to computer programs. McDowell claimed she was "training" the bot by correcting its perceived political bias during the chat.

Industry insiders suggest the current AI boom is primarily a race for "data capture" by hyperscalers like Google and Azure. While companies market AI stickers on products, the underlying technology is often standard machine learning, which remains expensive to run at scale. Gatekeepers within corporations are increasingly monitoring high LLM costs incurred by employee experimentation.

An industry insider from a major tech firm described current "AI" as merely advanced machine learning rather than true intelligence. The source argued that the technology is useful for sifting through structured data but remains a "parlor trick" in many consumer applications. The conversation also noted that AI music generators seem disproportionately tuned to produce country music.

A personal anecdote regarding a dress order highlights the limitations of current "agentic" AI in retail customer service. The automated bot initially refused a refund for an out-of-stock item, only for a human-like intervention to resolve the issue, suggesting that many AI bots lack the "empathy" required for effective retail.

Microsoft announces the Majorana 1 chip, claiming a breakthrough in error-resistant quantum computing. The hosts view this as a marketing "pivot" away from underperforming AI technologies, noting that while quantum computing promises to break Bitcoin encryption, commercial viability remains years away.

The success of TikTok's recommendation engine is compared to Bill Ziff's "special interest publishing" theory, where content and advertising are perfectly aligned to a specific mindset. A former ByteDance employee notes that the company has been training its machine learning algorithms for over a decade using global data sets. Concerns are raised that this massive data harvesting could eventually be utilized for sophisticated autonomous warfare and geopolitical influence.

During her PBS interview, Maria Ressa described how machine learning builds "clones" of users in a "motherlode database" to facilitate micro-targeting. She claimed this was used by Russia to impact 126 million Americans in 2016. The hosts mock her terminology, noting that "micro-targeting" is simply a modern term for targeted advertising that has existed for decades.

Jefferson County Public Schools in Kentucky experienced a "meltdown" after using an AI-driven mathematical model to redesign bus routes, leaving children stranded until 10 p.m. The district's $200,000 investment in the technology resulted in a total system failure and school closures. Additionally, a listener reports that a major U.S. bank is attempting to replace human customer service knowledge bases with AI, despite internal skepticism.

Vice President Kamala Harris spoke on the need for transparency and regulation in artificial intelligence, describing it as "two letters" representing machine learning. She emphasized the importance of controlling the information fed into AI systems to ensure legitimate decisions. Critics argue this focus on "correct information" is a move toward government censorship of algorithmic outputs.

Researchers have used artificial intelligence to identify a new antibiotic effective against the drug-resistant bacteria Acinetobacter baumannii. The hosts critique the media's personification of bacteria as "smart" and clarify that the process is actually high-speed machine learning rather than true intelligence. While successful in mice, the drug faces years of human clinical trials and the challenge of pharmaceutical profitability.

A "boots on the ground" report from an IT manager describes how software testing companies are rebranding basic tools as "AI-powered." Upon investigation, a supposedly advanced machine learning tool was revealed to be a simple spreadsheet containing standard formulas and macros.

MIT professor Max Tegmark is criticized for misdefining the Turing Test as simply "being able to talk like a human." The actual test requires a machine to be indistinguishable from a human in a blind exchange. Skeptics argue that ChatGPT's "flowery" and repetitive sentences do not yet pass this threshold and that the technology still struggles with basic factual correctness and multi-language translation.

Google CFO Ruth Porat's memo regarding "durable savings" and employee service cuts is deconstructed. The hosts mock the specific measures mentioned, such as reducing the frequency of muffins, removing personal staplers, and closing cafes on low-attendance days. They argue that Google's massive, unproven investment in AI (specifically Google Bard) is creating financial panic and forcing the company to rebalance its infrastructure spending.

The rise of large language models like ChatGPT has led to the phenomenon of "AI hallucinations," where the software confidently presents false information as fact. Educators are increasingly concerned about students using AI to write essays, though some argue this constitutes "machine learning bullcrap" rather than traditional plagiarism. Microsoft's $10 billion investment in OpenAI is scrutinized, with predictions that the technology may fail to meet its high commercial expectations.

The hosts discuss the potential for AI and machine learning to index the No Agenda archives by recognizing the "Clip of the Day" jingle waveform. They also provide an update on the show's donation model, noting that credits for producers will be read on the following Thursday show.

A historical review of the Plato computer learning system from the 1970s suggests that remote, computerized education is often a mediocre substitute for classroom instruction. Despite the availability of cheap hardware like the Raspberry Pi, critics argue that the fundamental flaws of automated teaching discovered decades ago remain relevant today.

Matthew Linton contributes $910.30 to become a knight, sharing a story of recovering from a near-fatal traumatic brain injury over the last decade. Now working in enterprise data management and outdoor pork farming, he credits the show with helping his cognitive recovery. He is knighted as Sir Matt Knight of the Casa Loma Wonders.

YouTube has introduced new requirements for creators to mark their content as "made for kids" or "not made for kids" following a $170 million FTC settlement over COPPA violations. The platform is shifting the legal burden of compliance onto individual creators, who must now navigate complex definitions of child-appealing content. Failure to comply could result in fines or the loss of monetization features.

Videos marked as "made for kids" on YouTube will lose key features starting in January, including comments, personalized advertising, and notification bells. The hosts review the FTC criteria for child-directed content, which includes the use of animated characters, toys, and simple songs. They argue that these changes will significantly decrease revenue for many creators and are part of a broader effort to "cull the herd" of non-brand-friendly content.

Internal Google documents reveal the use of "twiddlers," which are C++ snippets used to manually re-rank search results and down-rank specific URLs or news sources. The documents also discuss "Machine Learning Fairness," using examples like left-handed mugs to justify altering search results to meet diversity quotas. Evidence suggests Google maintains blacklists for sensitive topics like the Las Vegas shooting and applies negative weights to conservative outlets like Breitbart and Infowars.