Summary of What an AI-designed car looks like

Overview of The VergeCast — “What an AI-designed car looks like”

This episode focuses on how AI is changing car design and development, with a deep dive into how automakers may use AI to speed up everything from sketches and 3D modeling to wind-tunnel simulation and battery chemistry. The conversation then shifts to a fast-moving AI news roundup covering the Claude Code vs. Codex rivalry, OpenAI’s shifting public posture, Anthropic’s government and defense positioning, the fading usefulness of “AGI” as a term, and whether AI-driven layoffs are actually delivering real business gains. The episode also closes with a Slate truck update and a broader discussion of how AI hype is reshaping company behavior.

AI Is Changing How Cars Get Made

Tim Stevens explains that traditional car design is a long, iterative process that can take five to six years from concept to production.

How cars are normally designed

• Starts with sketches and product briefs
• Moves into digital drawings and 3D modeling
• Uses clay models — including full-size versions
• Then goes through:
  • wind-tunnel testing
  • engineering validation
  • virtual crash testing
  • software development and integration

Where AI fits in

AI is mostly being used to compress the timeline, not replace human designers:

• Turning sketches into 3D models in minutes instead of weeks
• Speeding up computational fluid dynamics simulations
• Helping with battery chemistry optimization
• Generating quick visualizations of what a concept car would look like in the real world

Why automakers care

• Faster development cycles
• Lower R&D costs
• Better ability to respond to changing consumer demand, regulations, and supply-chain conditions
• Potential to shorten the car development cycle from 5–6 years to around 3 years

What AI Could Change About Car Design

The discussion emphasizes that AI is more likely to improve efficiency than generate fully new car ideas on its own.

Potential benefits

• More affordable cars if development costs fall
• Faster iteration on design and engineering
• Better early-stage testing of whether a design actually works on the road
• More freedom to make bold design choices because bad ideas can be screened earlier

Risks and concerns

• Fewer entry-level tasks for new designers and engineers
• A weaker talent pipeline, since juniors often learn by doing low-level work
• Possible homogenization if AI just recombines existing design language
• The danger that companies will use AI to cut staff while claiming it’s only about “efficiency”

Big takeaway

AI may make cars faster and cheaper to develop, but it could also reduce the apprenticeship ladder that has historically trained the next generation of automotive designers.

Cars Are Becoming Software Products

A major point in the conversation is that modern cars are increasingly software-defined vehicles.

What that means

• Features like blinkers, horns, and safety systems are increasingly controlled by software
• Cars now involve huge software stacks, regular updates, and cybersecurity requirements
• Automakers need more software infrastructure than ever before

AI’s role in automotive software

AI could help with:

• documentation
• automated unit testing
• debugging
• update deployment
• patch management
• cybersecurity monitoring

Regulatory reality

AI won’t help a car pass:

• crash testing
• emissions testing
• safety standards

Those still require real-world validation.

Slate Truck Update

Tim also gives a quick update on the minimalist Slate truck.

What changed

• The truck’s original pitch relied heavily on a very low price
• Federal incentives have weakened, making the value proposition less compelling
• Slate recently raised $650 million
• The company brought in a new CEO, Peter Farisey, a former Amazon executive
• Production is still expected in Indiana before the end of the year

Why it’s still interesting

• The Slate truck remains a standout for its extreme minimalism
• It’s aimed at DIY buyers who want to customize with:
  • 3D-printed accessories
  • vinyl wraps
  • add-on features

Main uncertainty

At its current price, it may be harder to sell than it would have been under the original incentive structure.

AI Industry Roundup with Hayden Field

The second half of the episode is a wide-ranging AI news check-in.

Claude Code vs. Codex

• Claude Code remains the favorite among many developers
• Codex is gaining traction and seeing a user spike
• OpenAI is aggressively marketing Codex as its coding flagship
• Anthropic’s lead is still real, but it’s getting more public pushback and scrutiny

Bigger strategy shift

Both companies seem to be moving toward:

• developer-first tools
• workflow-specific products
• then broader “everything app” ambitions later

The money appears to be in business software and workflow automation, not consumer chatbots alone.

OpenAI’s current vibe

• Still messy, but slightly better than a month ago
• The company is trying to rebrand its public image
• It is pushing a more optimistic narrative about AI and prosperity
• That messaging does not seem especially persuasive to the public

Anthropic, the Pentagon, and “Any Lawful Use”

A major topic is the government’s growing relationship with AI vendors.

What happened

The Pentagon signed a deal with several AI companies, including:

• OpenAI
• Google
• Microsoft
• Amazon
• NVIDIA
• xAI
• Reflection

Notably, Anthropic was left out of this latest agreement.

Why it matters

The deal allows these tools for:

• “any lawful use”
• deployment on classified networks

That’s important because Anthropic had previously been the first company to get clearance for some classified network use.

Anthropic’s position

• Still relevant in government and defense circles
• Still viewed as having strong models
• But not currently locked into the newest defense agreement
• The company is actively trying to maintain and rebuild its government relationships

Mythos and AI Cybersecurity

The episode also touches on Anthropic’s cybersecurity-focused model, Mythos.

What it does

• Scans systems for vulnerabilities
• Crawls broadly rather than only checking for a pre-specified issue
• Can surface unexpected security holes

Why it matters

• Useful for defense and cybersecurity teams
• Powerful enough to raise concerns if deployed without guardrails
• Likely not unique for long, since competing models are moving quickly

Is AGI Dead?

Both hosts argue that “AGI” is increasingly a useless term.

Why the term is fading

• No one agrees on what it means
• Companies now prefer terms like:
  • human-centered AI
  • powerful AI
  • general-purpose AI

Why that’s useful

Getting away from “AGI” shifts attention from a mythical future moment to the real-world effects of AI today, including:

• labor displacement
• inequality
• harmful deployment
• vulnerable populations being affected first

AI, Job Loss, and the Layoff Question

A hotline question asks whether AI-driven layoffs are actually based on measurable ROI.

Main response

• There’s not much hard evidence that companies are doing rigorous ROI analysis
• Some firms likely overhired during the pandemic and are now correcting
• AI is often used as a convenient justification for cuts

What tends to happen

• Remaining employees are expected to use AI to absorb the work of laid-off staff
• This can lead to more overwork, not necessarily better productivity
• Companies may end up rehiring later after realizing they cut too deeply

Bottom line

The layoffs are probably a mix of:

• pandemic-era overhiring correction
• investor pressure
• genuine AI efficiency hopes
• public-facing hype and FOMO

Key Takeaways

• AI is already making car design and engineering faster, especially in early-stage modeling and simulation.
• The biggest risk in automotive AI may be the loss of junior-level training pathways.
• Cars are becoming software-heavy products, which creates huge opportunities for AI — and huge cybersecurity headaches.
• The Slate truck remains an interesting minimalist EV, but its economics are tougher now than when it launched.
• In AI, the big battle is increasingly about business workflows, not flashy chatbots.
• The term “AGI” is losing force, which may help shift attention back to the real-world impacts of AI.
• AI is often being used as a rationale for layoffs, but the actual productivity gains are still unclear.