The Neuron: AI Explained

What if neural networks are less like mysterious black boxes and more like systems we can inspect, debug, and eventually design with intention?

In this episode of The Neuron: AI Explained, Corey Noles and Grant Harvey talk with Eric Ho, Cofounder & CEO of Goodfire, an AI interpretability company working to understand what’s happening inside neural networks. Eric explains why models may contain meaningful internal structures — including features, representations, circuits, and curved manifolds — and how mapping those structures could make AI systems safer, more reliable, and more useful.

They discuss why models may “think in shapes,” how Goodfire uses AI to interpret other AI systems, what neural geometry can reveal about hallucinations and model behavior, and why interpretability could change how companies train and control their own models.

They also get into consciousness, robotics, multimodal models, the bitter lesson, and why Eric thinks more people should be looking under the hood of the most consequential technology of our time.

Subscribe to The Neuron for more grounded conversations about how AI actually works: https://www.theneuron.ai/

Sponsored by SAS AI Governance: Visit https://www.sas.com/en_us/solutions/ai/governance.html?utm_source=other&utm_medium=cpm&utm_campaign=sas-vpl-gbc-gps-global#

The Neuron Academy helps professionals build practical AI skills they can use right away, with lessons on prompting, workflows, and real workplace use cases. Check out https://theneuronacademy.com/ today!

AI agents are not just coming for laptops, inboxes, and office workflows. Samsara Cofounder and CTO John Bicket says they’re also coming for fleets, drivers, dispatchers, safety teams, and the physical infrastructure that keeps the world moving.

Recorded on-site at Samsara Beyond 2026 in Las Vegas, this conversation explores how Samsara is layering AI onto dash cams, telematics devices, asset trackers, vehicle data, and operational workflows. John explains how Agent Studio helps companies build useful AI tools without overwhelming frontline teams, why physical operations create different challenges than software-only environments, and how AI can help turn raw vehicle and safety data into coaching, risk reduction, maintenance insights, and real-time decisions.

Corey, Grant, and John also dig into AI ride-alongs, privacy concerns around in-vehicle cameras, why driver safety programs need to be framed around trust instead of surveillance, and what the day-to-day work of dispatchers, drivers, and maintenance leaders could look like three years from now.

Subscribe to The Neuron for more practical conversations about how AI is changing real businesses, not just demos.
https://www.theneuron.ai/

Join Grant Harvey and Corey Noles from The Neuron live on Thursday, July 9 at 10AM PT as we test OpenAI’s new GPT-5.6 model family in real time: Sol, Terra, and Luna.

Instead of just reading benchmark charts, we’re going hands-on with the use cases people should actually try first.

We’ll test:
🛠️ Building a working mini-app from a messy product idea
🧹 Refactoring a broken codebase without babysitting it
🧠 Using Ultra mode as a project manager with subagents
⚙️ Automating one annoying weekly workflow
🥊 Testing Sol vs. Terra vs. Luna on the same job
🔍 Running the same prompts head-to-head against Fable

Also: OpenAI just announced a 10AM PT livestream on July 9 for the next generation of ChatGPT Voice, reportedly featuring an upgraded bidirectional voice model, real-time capabilities, new voice samples, and voice orb colors tied to accent settings.

So we’ll also react to what OpenAI shows, explain what “bidirectional voice” actually means for normal users, and test where voice might finally become useful for work instead of feeling like you’re leaving a voicemail for a robot receptionist.

By the end, you’ll know what to try first, which model to use for which task, and whether GPT-5.6 or the new ChatGPT Voice actually changes your daily AI workflow.

📩 Subscribe to our newsletter: https://www.theneuron.ai

Most AI image tools give you a prompt box and a result. ComfyUI gives creators the pipeline underneath - the models, parameters, nodes, and repeatable workflows that can turn visual AI from a toy into production infrastructure.

In this episode of The Neuron, Corey Noles and Grant Harvey talk with Yannik Marek, co-founder and original creator of ComfyUI, about why node-based workflows matter, how open-source visual AI is moving into real creative production, and what teams gain when they can inspect, modify, and repeat every step of generation.

They cover how diffusion models work in plain English, why Comfy is useful for VFX, gaming, animation, e-commerce, and creative studios, and how Comfy balances open-source values with Cloud, API, and Enterprise products.

Yannik also shares practical hardware advice for running models locally, where open models are catching up fastest, and why the future of creative AI may depend less on a universal prompt box and more on visible, controllable workflows.

Try ComfyUI at comfy.org, and subscribe to The Neuron for more grounded conversations about AI in practice.
https://www.theneuron.ai/

The Neuron Academy helps professionals build practical AI skills they can use right away, with lessons on prompting, workflows, and real workplace use cases. Check out theneuronacademy.com today!

This week on The Neuron: AI Explained, Grant and Corey break down the strangest week in frontier AI so far: Fable 5 relaunching and getting yanked almost immediately, OpenAI’s GPT-5.6 rollout arriving in a weird half-launch state, and the bigger question underneath all of it.

What happens when governments can slow, restrict, or pause the most powerful AI systems right as the economy starts depending on them?

We’ll get into why these false starts matter beyond Silicon Valley drama. If U.S. labs keep getting caught between safety fears, export controls, and uneven release rules, China may get more room to catch up through open-source models, faster iteration, and fewer distribution bottlenecks. Meanwhile, businesses betting on AI have to plan for a world where the “best model” might disappear, degrade, get delayed, or become unavailable to half their stack overnight.

So this episode is our guide to navigating the current AI chop:

Why Fable 5’s relaunch and takedown became a warning shotWhat GPT-5.6’s limited launch says about frontier model accessHow restrictions could reshape the U.S. vs. China AI raceWhy every company needs an open-source backup strategyHow AI uncertainty could ripple into the broader economyWhat we’d do now as builders, buyers, workers, and AI-curious professionals

Join us for a strategic read on what’s changing, what’s fragile, and how to make smarter decisions while the AI ocean gets weird.