Musk bans resumes – asks for 3 bullet points instead

Musk bans resumes – asks for 3 bullet points instead

Comments 0 Comment

Resumes are becoming increasingly useless in the AI era. For top positions in his new chip design team, Musk now asks for something else:

Write 3 bullet points on the toughest technical problems you’ve solved

I like this approach. It cuts to the chase and really requires you to distill all those desperate skills you have into how you solved real problems. A resume is still required to apply for most other jobs at Tesla in the U.S.—with some positions even calling for an “evidence of excellence”. But Musk’s unconventional request follows a growing trend in skills-based hiring. Almost three-quarters of companies are using skills-based assessments during the hiring process.

AI Context storage – Fighting the amnesia loop

AI Context storage – Fighting the amnesia loop

Comments 0 Comment

Having big context windows is critical for LLM’s, but fitting a decent sized context on a GPU is rough work. Even when you do get a decent budget, the reality is that you’ll be running out of context more than a few times in an 8 hour session with a 120,000 or even 250,000 token context window.

It’s worse than that since research seems to indicate contexts start to degrade and lose coherence after you use up about 60% of your context budget. You can try /compress commands, but eventually things degrade enough you have to close and restart – which means you lose all that great context and have to retrain the LLM.

There’s a couple ways to fight this.

On Claude, you can persist sessions to external storage and resume sessions.

On Copilot CLI, the memory already exists. Copilot CLI writes every session to a local SQLite database — ~/.copilot/session-store.db. Desi Villanueva wrote a simple python script that will re-load that context if you use Copilot. Handy if you use it for Visual Studio Code.

Oregon in bottom half of growth since 2020

Oregon in bottom half of growth since 2020

Comments 0 Comment

Since 2020, Oregon has ranked in the bottom half of U.S. states for real GDP growth. While companies aren’t fleeing en masse, some are building new facilities and hiring additional workers elsewhere, citing higher costs, longer permitting timelines and a complicated path to growth in Oregon

Nearly a quarter of 400 Oregon businesses surveyed in a 2025 University of Oregon study said they had been contacted by out-of-state recruiters since 2021. Dutch Bros, the Grants Pass–founded coffee chain now valued in the billions, shifted 40% of its corporate staff to Phoenix last year.

Article

Adversarial Machine Learning Anti-drone attack

Adversarial Machine Learning Anti-drone attack

Comments 0 Comment

Researchers at UC Irvine have a more updated Adversarial Machine Learning attack that appears to work on drones.

By creating an umbrella with a specially painted top, the researchers were able to confuse the AI tracking and navigation of a DJI Mini 4 Pro drone and cause it to crash. The method is called the FlyTrap attack and is a visual pattern that performs a next-gen physical distance pulling (PDP) attack that works across multiple angles, even in motion in real settings.

The printed visual draws victim drones closer as its neural network tracking systems interpret the pattern to be the subject moving further away. As the drone approaches the umbrella, the pattern causes the targeting bounding box to continue shrinking – so the drone moves to get closer. Autonomous drones lured by the pattern can then easily be ensnared using a net gun, or further induced to crash to Earth.

None of this would stop a human-controlled drone of course (like we see in Ukraine). But as drone warfare becomes more and more autonomous – it would take down fully autonomous drones. Perhaps modern camo clothing needs to become anti-AI patterned more than traditional human visual camouflage. Or perhaps screens of this kind of camo can be used to protect bases or encampments.

Breakthrough in PS3 SPU emulation

Breakthrough in PS3 SPU emulation

Comments 0 Comment

The PS3 processor was paired with up to seven Synergistic Processing Units, each a 128-bit SIMD co-processor with its own 256KB of local store memory. It was notoriously hard to get working well in native games – simulating those SPUs has been even more complex and very difficult to make performant.

RPCS3 emulates SPU workloads by recompiling the original Cell instructions into native x86 code. It recently got a huge improvement by some new usage patterns discovered by elad335. He believe these could also speed up normal SIMD and CPU processes as well.

the Elad’s contribution identifies new patterns in how PS3 games use SPU instructions and implements more efficient recompilation for them.

LED floor

LED floor

Comments 0 Comment

People have done a lot with projection mapping onto floors, but what if you made a giant LED display as your floor? BMW paired with ASB GlassFloor to create a full-LED video sports floor. It was installed in Munich at BMW Park at the home venue of FC Bayern Basketball.

This does offer advantages over pure projection mapping since there are no issues with occlusion shadows (though that’s a problem largely solved already by using multiple projectors in multiple locations).

Personally, it seems terribly distracting if you’re trying to actually play on it or watch live play – but seems to have has possibilities for live entertainment/concert events or possibly new kinds of games. One could imagine games in which the floor changes configuration/boundaries. Color could be used to mark out 2 and 3 point regions. Perhaps colors could indicate different bonus scoring is in effect, or to pulse to count down the last 10 seconds of gameplay, or perhaps indicate a penalty is in effect (though this would be best only indicated as changes to the perimeter colors).

Using ChatGPT’s browser to run a solopreneurship

Using ChatGPT’s browser to run a solopreneurship

Comments 0 Comment

ChatGPT’s internet browser Atlas can now research, plan and execute tasks inside your workflow. Instead of using ChatGPT like a chatbot, others suggest using the browser in different ways.

ChatGPT browser use cases to scale work:

  • Content intelligence: Scan Reddit, Substack, and YouTube to build next week’s posting plan or feature lists from real demand
  • Tab chaos killer: Ask it what you were working on and rebuild your workflow from browsing context
  • Inbox cleanup: Auto-unsubscribe from dead senders, get a clean report of what changed
  • Content creation: Let the browser create breakout hooks, draft scripts and organize everything into a Google Doc automatically
  • Conversion boost: Audit your landing pages and get a ready-to-run A-B test plans
  • Smart purchasing: Compare tools, products, and software intelligently and quickly before you buy
First Meta, now the US Army

First Meta, now the US Army

Comments 0 Comment

After encouraging all their developers to use AI – including many companies that tracked how many tokens an employee used and used that for performance measurement and firing – many are now trying to hit the brakes.

First, Meta found out they were on track to spend $2B this year on tokens. Now the US Army is telling it’s members to hit the brakes as they’re running through their tokens faster than expected too.

Strange how people are taking advantage of the very things you told them to use with abandon or lose their jobs.

AI job swap: white collar workers ditching careers

AI job swap: white collar workers ditching careers

Comments 0 Comment

I was catching an Uber back home from the airport after a business trip this year when my 50-ish year old driver shared a shocking tale. He was a successful computer programmer for 7 years. He worked at a good company doing some very interesting work at a scientific company. His parents got ill, so he took a year off to care for them before they passed. When he tried to get back into programming, there were no jobs. He literally sent out hundreds and hundreds of resumes and applications to not get a single callback. He told me about cool coding projects he was working on in his spare time and projects he did for work. But it didn’t matter – he now has had to drive Uber the last year to even make ends meet. He doesn’t think he’ll ever get back into computer programming again.

Another friend of mine recently got out of game development because the industry was pretty toxic (something I can confirm) and was also imploding due to the worst layoffs seen in decades. She is now training to become a physical therapist. She’ll make half what she did before, but is already happier and feels she is doing more good for the world than game programming ever did.

The change is dramatic and it’s real. Everything can’t be blamed on AI. There was an amazing amount of over-hiring during Covid that is now being unwound. But it is clear that is is impacting people – especially entry level workers that can’t even get in and people over 50. My Uber driver’s story isn’t the only one – and workers in Europe are seeing the same patterns – even outside of software.

Jacqueline Bowman consistently got paid freelance writing work (marketing/journalism) every day for years. In 2024, that all started to dry up as more and more of her clients started talking about AI. Soon she was reduced to editing work for about half the pay – despite needing to take even more time due to fact-checking everything the AI drafts were generating. By Jan 2025, she could no longer pay her bills. She’s now switching to psychology and hoping it will be more AI proof.

Janet Feenstra was a freelance academic editor in 2013 – but now turned baker in Malmö, Sweden at 52 years old. She realized the writing on the wall when she saw AI could do the kind of english language editing she was doing for non-native speakers. She had to give up her home, retrain for a year, and now makes much less and lives in a small flat.

Richard was a occupational health and safety professional that worked his way up the career ladder for 15 years. He saw more and more policies and safety systems done by AI. He sees need for a few “highly specialised managers. The rest of it’s probably going to go.” He jumped ship and is becoming an electrical engineer. He also took ‘a huge cut’ financially.

Even students and youth are starting to switch paths. Paola Adeitan was studying law, but realize friends of hers were struggling to find entry jobs caused by the rise of AI use in law firms. Another 23 year old left his pursuit of a geography degree for a trades job.

2023 report on the impact of AI on jobs and training in the UK by the Department for Education outlined that many clerical, finance, law, and business management jobs are very vulnerable to AI. The most significant AI-caused declines in employment and wages will be in jobs like software engineering and management consultancy, according to a King’s College London study.

What comes next is going to be very disruptive – very much like the revolution of the internet. Who comes out the winners is rarely who went into it that way. It’s good advice to pay very close attention to what the trends are in your industry – because AI is already affecting your field.