Logixly Live Unsupervised Learning Podcast

Anthropic's latest AI model release is making waves, but is the company really pushing AI forward as fast as it could?

In Episode 14 of Unsupervised Learning, the NLP Logix team breaks down Anthropic's latest model advancements, the growing competition between Claude, OpenAI, and Google, and the debate over whether AI companies are deliberately slowing development.

From an AI experiment involving more than 1,000 sub-agents to the question of whether benchmark improvements translate into real-world value, the conversation explores what progress in artificial intelligence actually means.

In this episode, we discuss:

• Anthropic's latest model release: How Claude is performing against competing models from OpenAI and Google.
• The AI benchmark race: Why higher benchmark scores don't always translate into better everyday experiences.
• 1,000+ AI agents working together: Anthropic's reported experiment using autonomous agents to build an operating system prototype.
• Is Anthropic slowing down AI development? The debate around AI safety, responsible development, and competition between frontier AI companies.
• Measuring real-world AI performance: Why evaluating AI on complex business problems is more challenging than comparing leaderboard rankings.
• The Jacksonville RFP investigation: An update on the team's ongoing effort to obtain public procurement documents and the challenges involved.

As AI capabilities continue to advance, one question remains: Are we measuring progress by what AI can accomplish, or by how much value it actually delivers?

New episodes of Unsupervised Learning every Thursday! Subscribe and turn on notifications to stay up to date on the latest conversations in artificial intelligence.

When AI gives you an answer, who—or what—is influencing it?

In Episode 13 of Unsupervised Learning, the NLP Logix team explores how AI-generated answers are changing online search, advertising, and the way we consume information. As users move from clicking links to accepting summaries, what happens to the businesses and creators behind that content?

The conversation moves from advertising designed for AI agents to model steering and hidden instructions in documents, raising questions about transparency, trust, and human oversight.

In this episode, we discuss:
• How AI summaries are changing search habits and website traffic.
• What advertising could look like when AI agents become the audience.
• The difference between sponsored content and influencing an AI response.
• How hidden instructions in documents can attempt to steer AI outputs.
• Why human review remains important when using AI in consequential workflows.

Would you trust an AI recommendation if you knew it had been influenced by paid advertising? Share your thoughts in the comments.

MEET THE HOSTS:

Ted Willich:

/ Ted Willich | LinkedIn

Matt Berseth:
/ Matt Berseth | LinkedIn

Katie Bakewell:
/ Katie Bakewell | LinkedIn

CONNECT WITH NLP LOGIX:

Learn more about NLP Logix:
https://nlplogix.com/

Follow us on LinkedIn:
/ NLP Logix: Home | LinkedIn

Follow us on Instagram:
/ Instagram

New episodes of Unsupervised Learning every Wednesday! Subscribe and turn on notifications to stay up to date on the latest conversations in artificial intelligence.

Is AI making us smarter, or are we slowly losing the skills that make us human?

As artificial intelligence becomes part of our everyday lives, a new concern is emerging: AI de-skilling. When we rely on AI to complete tasks, are we becoming more productive, or are we losing the ability to do those things ourselves?

In this episode of Logixly Live Presents: Unsupervised Learning, Matt Berseth and Katie Bakewell debate whether AI is weakening human expertise or creating opportunities to develop entirely new skills.

From research on AI-assisted medical procedures to students using ChatGPT, hackathons, and the future of software development, they explore what happens when technology starts doing the work we once had to learn ourselves.

In this episode:

• What AI de-skilling means and why people are talking about it
• Research raising questions about AI dependence and human expertise
• Whether AI is limiting creativity and critical thinking
• How students can use ChatGPT to learn rather than simply get answers
• Why re-skilling may be more important than de-skilling
• Could voice-based AI eventually replace the keyboard?
• What skills will matter most in an AI-powered future?

The big question: Should we be worried about the skills AI is replacing, or focused on the new skills it allows us to develop?

Join the conversation in the comments! Do you think AI is making us more capable or more dependent?

RESOURCES MENTIONED:

Research Study: AI-Assisted Colonoscopies and the Risk of De-Skilling:
https://www.thelancet.com/journals/langas/article/PIIS2468-1253(25)00133-5/abstract

MEET THE HOSTS:

Matt Berseth:
  / mattberseth 

Katie Bakewell:
  / katiebakewell 

CONNECT WITH NLP LOGIX:

Learn more about NLP Logix:
https://nlplogix.com/

Explore the Unsupervised Learning Podcast:
https://nlplogix.com/insight/logixly-...

Follow NLP Logix on LinkedIn:
  / nlp-logix 

Subscribe on YouTube:
   / @nlplogix 

New episodes of Unsupervised Learning every Wednesday! Subscribe and turn on notifications to stay up to date on the latest conversations in artificial intelligence.

AI feels like it’s everywhere and nowhere at the same time. One person is still figuring out how to ask for a better email draft, while another is building agentic workflows that quietly reshape an entire operating model. That tension is why Andrej Karpathy’s AI adoption funnel is so useful, and we unpack it from a real-world angle: where people actually are on the spectrum and how that shapes what they believe about “AI hype.”

We also get practical about what stops adoption cold. Buying licenses for ChatGPT, Claude, copilots, or internal assistants doesn’t automatically create capability, and we explain why “the license is not a capability” should be the mantra for enterprise AI rollout. Special guest Ashton Gideon, Senior UX/UI Designer at NLP Logix, shares the experience design lens: sit with people, watch the messy workflow details, listen for hesitation, and design the smallest changes that create an “aha” moment. That’s how AI moves from novelty to habit.

Then we tackle the UX problem everyone feels but rarely names: today’s chat interfaces and constant mode decisions create friction, and the output is starting to overwhelm the humans who have to review it. We talk about proactive assistants, the challenge of managing hundreds of hours of generated work, and why the next layer is orchestration, not more tokens. If you care about AI adoption, AI user experience, prompt strategy, and building the right things (not just shipping fast), you’ll get a lot out of this one.

Connect with the Unsupervised Learning Team!

• Matt Berseth, Co-Founder & Chief AI Officer at NLP Logix
• Ted Willich, Co-Founder and CEO at NLP Logix
• Ashton Gideon, Senior UX/UI Designer at NLP Logix

Send our team a message here to discuss more topics from this week's episode or a topic you'd like them to dive into for a future episode!

The 2026 State of AI Report is here, but what does it reveal about what's actually working in artificial intelligence?

In Episode 10 of Unsupervised Learning, Matt Berseth and Katie Bakewell of NLP Logix break down some of the most interesting findings from the latest State of AI Report, exploring the evolution of AI models, the importance of real-world applications, and where AI still has room to improve.

From OpenAI and Anthropic's competing approaches to innovation to the role of AI in scientific research and decision-making, this conversation goes beyond the headlines to explore what these developments mean for the future of AI.

In this episode:

• OpenAI vs. Anthropic: Different approaches to AI innovation
• Why better AI performance isn't always about better models
• The role of prompts, tools, and supporting systems in AI performance
• AI-generated research and the challenges of scientific discovery
• Why AI predictions don't always translate into real-world success

RESOURCES MENTIONED

2026 State of AI Report by Nathan Benaich of Air Street Capital:
https://www.stateof.ai/State-of-AI-Report-2026.pdf

Nathan Benaich on LinkedIn:
https://www.linkedin.com/in/nathanbenaich/

MEET THE HOSTS

Matt Berseth:
https://www.linkedin.com/in/mattberseth/

Katie Bakewell:
https://www.linkedin.com/in/katiebakewell/

CONNECT WITH NLP LOGIX

Learn more about NLP Logix:
https://nlplogix.com/

Explore the Unsupervised Learning Podcast:
https://nlplogix.com/insight/logixly-live/

Follow NLP Logix on LinkedIn:
https://www.linkedin.com/company/nlp-logix

Subscribe on YouTube:
https://www.youtube.com/@NLPLogix

New episodes of Unsupervised Learning every Wednesday! Subscribe and turn on notifications to stay up to date on the latest conversations in artificial intelligence.

#UnsupervisedLearning #StateOfAI #ArtificialIntelligence

OpenAI’s latest math results make us wonder what “progress” really means when a model can generate proofs quickly but still struggles with the messy human work of deploying AI inside a business. 

We talk through why mathematicians care about elegant proofs, why counterexamples are a perfect job for computers, and why enterprise AI still needs people who can integrate, validate, and adapt systems over time. 

• OpenAI’s reported jump in math output and why the pace feels exponential 
• What it means for us to be an OpenAI select partner and what the application emphasizes 
• The Wired-reported tension around pacing research releases and mathematicians’ concerns 
• Beauty in mathematics versus “a proof that works” and the Erdős number detour 
• Why counterexamples and bound improvements play to machine strengths 
• Human factors in business change management and why emotional intelligence still matters 
• Data drift, monitoring, and how bad inputs and shifting conditions break systems 

Hit us up on LinkedIn. Keep it up if there’s something you want us to cover specifically.

Matt Berseth

Ted Willich

Katie Bakewell

NLP Logix

Unsupervised Learning Podcast 

Follow us on LinkedIn and YouTube

Things the team talked about in the podcast:

• OpenAI Sharing AI Profess in Mathematics: https://openai.com/index/sharing-ai-progress-in-mathematics/
• OpenAI is Frustrating Mathematicians Again: https://www.wired.com/story/openai-is-pissing-off-a-bunch-of-mathematicians-again/
• Git Hub OpenAI Math Repository: https://github.com/openai/math

We break down the reported details behind Anthropic’s private S-1 and why an Anthropic IPO conversation quickly turns into a debate about switching costs, and the brutal economics of frontier AI. We also zoom out into what actually drives business value with generative AI, from narrow models and unbundled tools to the real-world quality-of-life gains that don’t make headlines.

• Anthropic’s reported revenue ramp and what “$65B run rate” implies
• Putting AI revenue scale next to Amazon, Walmart, and Apple
• Frontier models versus small language models for narrow workflows
• Unbundling AI into task-specific tools like decisioning and coding
• Why Claude wins mindshare with developers and why that matters
• Switching LLM providers compared with switching internet or streaming
• AI as a utility, regulation, and the tradeoffs of standardization
• Is AI more important than electricity or the internet over time
• Practical optimism on productivity and quality of life, including healthcare documentation

Things the team talked about in the podcast:

Anthropic's IPO Prospectus 

All-In Podcast

We want to hear from you:

Matt Berseth

Ted Willich

Katie Bakewell

NLP Logix

Unsupervised Learning Podcast 

Follow us on LinkedIn and YouTube

We dig into JEV, a decision-focused AI model that trades open-ended chat for constrained, software-ready outputs. We talk through why that shift matters for enterprise automation, cost, speed, calibration, and trust when AI starts choosing the next action in a workflow. 

• Why JEV is positioned as a decision model rather than a chatbot 
• The split between human assistants and automated business processes 
• Unbundling frontier LLM capabilities into specialized models 
• Why light models feel risky for enterprise classification accuracy 
• Cost, quality, and speed trade-offs when AI runs at scale 
• What confidence scores and probabilities should mean in production 
• Calibration differences between general LLMs and task-trained models 
• How decision models can bridge toward fine-tuning on proprietary data 

Everyone has an opinion about the next frontier model. That is why we’re zeroing in on the rise of the forward deployed engineer: the person who embeds with a customer, learns the operation up close, adapts the tech to fit real constraints, and stays long enough to make sure the solution actually gets used.

We unpack where the term comes from (including Palantir’s view of forward deployed engineers and the “devs vs deltas” split), and why the role is suddenly showing up in hiring conversations like “data scientist” did years ago. We also wrestle with the weird split in the market: The debate on AGI and whether progress should slow down, while those same companies build services and deployment teams because the biggest bottleneck is not intelligence, it is execution. For most organizations, a good model deployed beats a perfect model on a shelf.

Along the way, we get concrete about what last mile AI deployment really takes: workflow mapping, change management, domain learning, and tight iteration with the people doing the job. We share field stories that highlight why specs miss critical context, from financial trading floors to dangerous manufacturing environments, plus a cautionary tale about “snake vision” and why cool technology can still be the wrong solution.

If you’re building enterprise AI, leading digital transformation, or trying to ship automation that sticks, this one will sharpen how you think about roles, loops, and outcomes. 

We unpack why recent AI headlines feel less like progress and more like a trust breakdown, from Open AI's GPT-6 Astra becoming less transparent to big claims about hitting AGI (artificial general intelligence). We also dig into the Millennium Prize math controversy and what it suggests about telemetry, governance, and whether anyone should hand over their workflow data to an AI provider.

• GPT-6 Astra getting less transparent and why traceability matters
• What AGI should mean versus passing tests
• Why “AGI is here” should feel world-changing
• The Millennium Prize problems and why they matter
• Questions about prompts, drafts, and telemetry influencing outcomes
• Trust, audits, and governance as the real battleground
• Marketing demos that don’t match how people use AI