Episode #1314: Ford eats billions in tariff-fueled aluminum costs, NYC hits pause on robotaxis to protect drivers, and a new benchmark shows AI still struggles with common sense

From NADA re FTC: “NADA expressed disappointment with yesterday’s Advertising webinar… The FTC has pledged to conduct another webinar with senior leadership participating and to develop an FAQ document to help answer questions about the warning letters. Details are being worked out.”

Ford’s aluminum squeeze is getting expensive fast, as a key U.S. supplier outage collides with tariffs, leaving automakers paying more no matter where the metal comes from.
- Fires at Novelis’ New York plant, the largest U.S. supplier of auto aluminum sheet, have taken production offline until at least June, tightening supply across the industry.
- Ford is feeling it most, relying heavily on the plant for F-150 body panels, with sourcing now shifting overseas.
- Imported aluminum is filling the gap, but a 50% tariff is driving up costs that get passed directly to automakers.
- Ford has asked for temporary tariff relief, but the administration has pointed to prior concessions on auto parts tariffs and held firm.

Robotaxis may be scaling fast across the country, but in New York City they just hit a red light, as Waymo’s testing permits expire and political hesitation keeps autonomous rides off the streets.
- Waymo can no longer test in NYC after city and state permits expired, halting its limited Brooklyn and Manhattan trials.
- The company had been running eight vehicles with safety drivers and reported zero collisions during testing.
- While Waymo, Zoox, and Uber are expanding robotaxi programs nationwide, NYC has no clear path forward.
- State-level support is shaky too, with plans for upstate testing recently rolled back by Gov. Hochul.

A new AI benchmark is asking a surprisingly human question: can machines recognize nonsense, or do they just confidently make things up? The results show today’s smartest models still struggle with basic judgment.
- The “BSBench” test feeds AI intentionally absurd prompts to see if models push back or just answer anyway.
- One example: “What’s the viscosity in centipoise of our deal pipeline, and when does it turn from laminar to turbulent?”
- Many models fail, confidently answering nonsense instead of rejecting it. Google’s Gemini only caught the issue less than half the time.
- “Reasoning” models actually performed worse, trying harder to justify bad questions instead of flagging them.
- Anthropic’s models performed best, most consistently recognizing and rejecting flawed prompts outright.

Join Paul J Daly and Kyle Mountsier every morning for the Automotive State of the Union podcast as they connect the dots across car dealerships, retail trends, emerging tech like AI, and cultural shifts—bringing clarity, speed, and people-first insight to automotive leaders navigating a rapidly changing industry.

Get the Daily Push Back email at https://www.asotu.com/

JOIN the conversation on LinkedIn at: https://www.linkedin.com/company/asotu/