First Take

4th Shift
New tools and technologies are coming out every week that have an impact on our businesses, careers, and personal lives. In this issue, we identify some AI sense and world model embodiment advances toward AGI as well as some new robots that are out there. We also look at free gasoline from air and a supposed Anti-gravity engine. One of these is a real product coming soon and the other is about as likely as passing a $3 bill at a bank. I know which is which and you will too if you read my take on it in this issue of The Shift Register. Additionally, some new AI driven tools like NotebookLM from Google are making marketing departments super happy by providing for some instant podcast style content generation based on their adcopy. Meanwhile, here at The Shift Register I'm using the same tool to try and expand our own audience base. You can LISTEN to the last issue here.
Some of these AI tools are truly amazing in what they can accomplish without hallucinations. Keep in mind that this specific tool, while reducing content creation costs and effort levels, is absolutely killing some job opportunities. Not that I would hire any pod-casters to summarize our issues, but folk with actual marketing budgets might have hired out that kind of work previously. Why would they do so now? What's your take on this particular SHIFT? Comment on our Facebook page to let us know. Speaking of which, you can click to get Issue #4 in podcast form courtesy of NotebookLM right now if you would rather listen than read it.
AI

AI learns how vision and sound are connected, without human intervention | MIT News | Massachusetts Institute of Technology
A new machine-learning model learns to pinpoint exactly where a particular sound occurs in a video clip without the need for human intervention. The model could have applications in areas like journalism and film production or education and training.
These multiple input processing type systems are slowly getting closer to an all pervasive human-like embodiment that could result in AIs capable of understanding and interacting with the real world much like we do.

Brain-Inspired AI Learns To See Like Humans in Stunning Vision Breakthrough
A team of researchers from the Institute for Basic Science (IBS), Yonsei University, and the Max Planck Institute has developed a new artificial intelligence (AI) technique that brings machine vision closer to the way the human brain processes visual information.
My take is that the closer we align AI sense processing to human sense processing, the closer we get to a general AGI embodied in a humanoid robot. No idea if that will be a good thing, but it is apparently a thing we are going to find out in the near future.

Google’s 'world-model' bet: building the AI operating layer before Microsoft captures the UI | VentureBeat
Google doubles down on its ‘world-model’ vision, racing to build an AI operating layer to drive a universal personal assistant with Gemini. Even as Microsoft moves to capture the enterprise UI. Here's what's at stake.
My take is that is another step in trying to gain AGI with an understanding of the real world. Google in particular has said they can hit that mark by 2030.

Until NotebookLM, I never believed AI could be this game-changing for productivity
Slowly but surely, AI is beginning to reshape how we work and even spend our day-to-day lives. Despite that, many still find it difficult to accept that it's here to stay. But the truth is, the further you try to run from artificial intelligence, the quicker you'll get left behind.
My Take is that the new AI driven tools out there are driving some unique content creation in unforeseen ways. While this tool could have helped with my PhD work, I have found the discussion creation tool possibly useful to expand The Shift Register's outreach. Some of my favorite nerds spend more time watching YouTube or listening to podcasts than reading and I can reach them with NotebookLM's discussion tool.

Anthropic CEO says AI could cause up to 20% unemployment within five years, wipe out half of all entry-level white collar jobs | Tom's Hardware
AI is seemingly the train that cannot be stopped.
My take is that Anthropic is probably giving us pretty realistic numbers here. 20% is a HUGE number. People need to get their job seeking game to the next level and fast.

AI revolt: New ChatGPT model refuses to shut down when instructed | The Independent
OpenAI’s o3 model raises AI safety fears after sabotaging commands for its own self-preservation.
My take is that these kinds of issues seem to be increasing in frequency as the systems get smarter. Now's a good time to take another hard look at what we are doing and ensure we have adequate non-rewrite-able hardware controls in place. George Carlin solved this problem decades ago by giving all robots a 6' power cord to keep them under our thumbs. ;-)
Emerging Tech

Fridge-sized machine produces clean gasoline from air and CO2
Gasoline from air became reality at Aircela’s NYC demo, where a working machine produced fossil-free fuel in real time.
My take is that I want one of these machines at my house. Take my money!!!!

US engineers bend quantum rules to turn entangled light into 3D hologram
Brown students use quantum entanglement and dual wavelengths to create holograms without infrared cameras.
My take is that this should offer some improved medical imaging options while reducing costs.
News

Revolutionary anti-gravity invention stuns scientists by defying physics live - Jason Deegan
A bold new propulsion technology claiming to defy fundamental physics has ignited both awe and skepticism among researchers. Promising to revolutionize space travel by generating thrust ... Continue Reading →
My take is that this is utter and complete nonsense. The news part is that it is bogus and keeps getting published as real news. This guy has been making these unproven claims for decades and eating every investment dollar he can ingest. IF he'd publish his results in a fashion that permitted 3rd party testing, or even allowed 3rd party testing I might say WOW! Barring those outcomes, it's just garbage news that cycles around every time he gets someone new to listen to his stories. I only ever recommend real products with potential and while I still want that space elevator, I KNOW there isn't a solved materials option that permits its creation as yet and the company that is working the problem has been throwing away investor dollars for 2 decades now (that I've been aware of the) with zero progress beyond millimeter length carbon nanotubes.
Robotics

These robot cats have glowing eyes and artificial heartbeats – and could help reduce stress in children | Robots | The Guardian
Lifesize interactive pets touring libraries in Blue Mountains are also touted as being able to comfort people living with anxiety or dementia.
My take is that we have plenty of real pets in need of adoption, but if you can't care for a pet, a robot pet might do in a pinch.

Humanoid robots fight in Chinese boxing competition
The two humanoid robots traded punches while fans watched on, in a world-first humanoid robot fighting event.
My take is that I thought the US had the most advanced robot fighting arena matches until I saw this. Wait until they start wielding flame throwers and saw blades. Then I'll really be impressed.

Atlas robot gets superhuman 3D vision in major Boston Dynamics upgrade
Atlas can now see, localize, and manipulate objects with millimeter accuracy thanks to Boston Dynamics’ major vision system upgrade.
My take is that when they get the unified foundation model running, we will really start to see some emergent capabilities.

Delivery bots learn to tackle those last 100 steps
Swiss robotics company Rivr is partnering with Veho, a gig-based delivery company, to test its wheel-legged robot in Austin, Texas.
My take is that despite the article saying no delivery drivers would be replaced, they are obviously looking for an automated way to place packages on doorsteps. Delivery driving is a different piece of automation being worked on elsewhere. End to end logistics work is ALL fair game for today's automation efforts. Picking, sorting, loading, unloading and the drive have all received attention. This is exactly what it says in the title, "Those last 100 steps".
Open Source

Docling: An open-source toolkit for advanced document processing | InfoWorld
Docling uses state-of-the-art models for layout analysis and table structure recognition to transform unstructured documents into formats readily consumable by modern AI systems.
My take is that this pre-processing tool can help present your AI systems with a standardized contextual understanding of complex documents. This will help make things with mixed tables, charts, and photos more comprehensible to LLM based AI systems.
Security

26th May – Threat Intelligence Report - Check Point Research
For the latest discoveries in cyber research for the week of 26th May, please download our Threat Intelligence Bulletin. TOP ATTACKS AND BREACHES Cellcom, a Wisconsin-based wireless provider, has been impacted by a cyberattack that resulted in widespread outages of voice and SMS services beginning on May 14, 2025. The incident disrupted communication for customers […]

Remote Zero-day Linux Kernel Flaw Discovered Using AI
Picture this: you're staring at your trusted Linux system, a network of processes handling everything from file sharing to user requests. It's smooth, reliable, and—if you're confident in your setup—secure. But then comes word of yet another critical vulnerability. CVE-2025-37899
My take is that while this sounds super scary, there's a bit more texture to it. First, yes you can use AI to discover flaws. Second, this particular flaw is about as bad as they can get, but only a small subset of Linux machines SERVING SMB shares using kernel based code are impacted. Turn off SMB file sharing on your Linux machine and you are safe. Run SAMBA in user-land and you are again safe.

9,000 Asus routers compromised by botnet attack and persistent SSH backdoor that even firmware updates can't fix | Tom's Hardware
The backdoor evades detection and survives firmware updates.
My take is that SOHO routers without security issues are getting very difficult to find. I can't even make a recommendation for fear that my selection will be targeted due to the publicity. Good luck out there!
Final Take

Last Looks
As AI and robotics merge into an embodied intelligence operating in the real world and performing real work, where does that leave those of us unable to design, repair, program, or maintain such systems? There are really only a couple of options. One of these is universal basic income coupled with self actualized work or volunteer activities. The other is augmentation and direct competition. We currently lack the technology to even do the augmentation work to any meaningful level, but it's possible we get there with AI's help. Assuming both of these options are available in the near future, which path would you take? Why would you choose that over the other? Hit us on our Facebook page or take our poll and let us know.