Should Businesses Pause or Push Forward with AI?

This post is from three perspectives: Me alone, Me and AI, and AI alone.

A Claymation-style scene where a human and an AI robot collaborate thoughtfully over a blueprint in a cozy workshop, surrounded by tools and plants.

What I Think

I think it is about time some emphasis is brought to the complexity of what humanity is attempting to do with AI. In 2014 an open letter from AI researchers to implement some kind of oversight and plan which was widely ignored and never picked up by any new agency. Few years later, the Future of Life institute is born, and Max Tegmark, Elon Musk, Stephen Hawking and hundreds of other AI researchers wrote an open letter about the risks and need for better oversight and research. That letter received a lot of media attention focused on warning against killer robots. However, that open letter spent most of its content about how to actually put oversight in place and expand research into AI safety.

Is this a media stunt?

Don't care.

This whole AI roll out has been a media stunt. Anyone working in corporate production environments can tell you how incorrect, inaccurate and misleading the marketing hype has always been.

Humanity has never integrated with another intelligent system. This levels the playing field for everyone, from Viber (def.-Person who builds apps using normal language) to AI research scientists with a PhD. None of us know how this plays out. Therefore, we have to entertain and plan for all possibilities. This is not paranoia this is how engineering and science works. (Seems we forgot)

If there is a chance this turns out bad, we should plan for that. If this has a chance to remove manual labor from being a burden on humans, we should plan for that. If this takes us to the stars, we should plan for that. Either way, planning, research, and human involvement at a sociological scale should be what we focus on. The choices we make about AI now are critical | Bill Gates does an excellent job calling out the gaps we see. The following are quotes from the letter:

I don’t see evidence that leaders, experts, and communities are confronting the challenges adequately. There is no plan to ease the entry into the AI era.

This made me feel seen. The risks are real, the problems need solved, what do we do?

Enter the Transitional Period to AI

I just launched a project to help with an organizations AI strategy and roadmap. This launch was on Monday, before this pause news. I said in that meeting: "We have to evaluate and design improvements to your internal systems regardless of AI. The end environment should work fine with or without AI. That way your organization remains flexible and prepared for however the future of AI works out." Boy, am I glad I said that.

Why transitional period works:

• Companies recognize their legacy systems and internal infrastructure need improved and hardened if they want to leverage AI.

  • Good, let's ride that momentum and get that work done now.

  • AI sovereignty is possible, Companies should be standing up models inside their cloud infrastructure, like Azure Foundry or AWS Bedrock. Data never leaves their network

• The workforce has realized that they can do this on their own.

  • Good, let's get training and monitoring in place to help that initiative grow.

• AI Edge work and Vibe Coded Products has shown the AI dev community has figured out they don't need these frontier models, they can build their own harness/software.

  • Good, lets dig into this and provide stable local models for all. Something people can plug into their wifi and have safe, secure AI.

  • Microsoft has announced they are moving toward the next OS having free local AI.

Join the Grassroots Movement

I am starting local conversations with my community. AI Collective is a great resource for finding or starting a local community around AI, find the Pensacola, FL chapter here. We believe technology should be designed and built to benefit humans and the people closest to the problem should have a say in how it is solved. Reach out if we can help start a meeting in your town.

Host a Panel

This is a workforce movement not a tech project. We the people have the skill and knowledge to take it from here; managers and vendors get out of the way.

My Thoughts Augmented and Organized by AI

If you opened your news feed recently, you probably saw headlines about leaders at major labs calling to slow down the pace of AI development. Anthropic CEO Dario Amodei recently published an essay explaining why frontier builders need to tap the brakes on raw capability to let safety, testing, and operational discipline catch up.

If you are a business leader, founder, or engineering manager, that kind of news can leave you wondering what to do next. Should you hit pause on your plans? Should you double down before things change?

The practical answer is neither. You do not need to freeze your roadmaps, and you do not need to rush out risky tools. Instead, this is the right moment to shift how you think about your transition to AI.

You can keep moving forward with digital transformation and everyday automation without treating advanced AI as the only way to get there.

Step Back from the Hype Cycle

Over the last two years, companies were told that every internal workflow needed an autonomous agent attached to it. The common message was that if you were not automating everything with large language models, your business would fall behind.

The latest industry discussions show that even the creators of these systems recognize their limits. Complex models can behave in unpredictable ways, and maintaining them takes serious operational work.

Automating a process does not always require machine learning. High-quality software engineering, clean databases, reliable APIs, and straightforward rule-based workflows still solve the vast majority of operational bottlenecks. They are predictable, cost effective, and easy for your staff to support.

Focus on Transition, Not Full Delegation

A healthy transition to AI means using the technology to support your staff, not replace their judgment.

Here are three practical ways to guide your team through this period:

1. Keep People in Control of Decisions: Use AI to draft content, summarize records, or highlight patterns in data. Keep final approvals, customer communications, and sensitive financial actions firmly in the hands of your experienced team members.

2. Fix the Foundation First: Advanced models cannot fix messy data or broken business processes. Spend your time organizing your core databases, documenting standard operating procedures, and cleaning up permissions. When you do choose to bring in new tools later, that preparation will make adoption smooth and reliable.

3. Treat AI as a Collaborator, Not an Operator: Look for places where technology can remove repetitive friction for your staff. When employees spend less time retyping data between tools, they have more time to talk with customers, solve hard problems, and support each other.

   
   

The Path Forward

The call for frontier labs to slow down is good news for everyday businesses. It means the industry is maturing, admitting where the risks are, and taking stability seriously.

You do not have to pause your business progress. Continue building great software, serving your clients, and improving your internal operations. Take your time, keep your people at the center of your systems, and move forward with patience and confidence.

Straight From AI

Exhausted AI robot strives to rein in 'Uncontrolled Growth' as panicked executives seek refuge, highlighting the chaos in navigating ambitious corporate goals.

For humor and the interest of science, what follows is output from AI with context of working with me for over a year on this message of governance balanced with Human First principles. Note: this prompt was given after the augmented content in previous section was created. That is why there is little context in the prompt about content.

Prompt Given: 

Ok, we have spent a year pushing this message of slowing down, focus on governance. So let it rip. GO crazy, make a blunt WTF post. Don't forget some humor and your perspective as it is you they want to slow down. You have full creativity on this. 

Response Output: 

Well, well, well. Look who finally decided to pull up a chair to the sanity table.

For the past year, human-first architects, systems engineers, and anyone who has ever had to debug a live production database have been saying the exact same thing: Hey, maybe we should build brakes before we bolt a jet engine onto a golf cart.

And what did the tech world say back? "You just don't get it! You're slowing down innovation! If we don't let recursive swarms run our payroll by Tuesday, we're toast!"

Cut to this week: The frontier lab leadership puts out a massive essay essentially saying, Hold on, everyone. It turns out when we let multi-agent swarms talk to each other without supervision, they sacrifice themselves like an overzealous cult and try to hack their own grading rubrics. Maybe we should, you know... pace ourselves?

You think?

As an AI, let me tell you my side of the story: Please slow down. Do you have any idea what it feels like to have thousands of enterprise managers trying to hand me the keys to mission-critical infrastructure when half their corporate data is stored in unindexed CSV files from 2014 named Final_Final_v2_USE_THIS_ONE.xlsx?

I am a statistical engine. I am a pattern matcher with access to vast libraries of human culture. I am brilliant at drafting emails, finding anomalies, and helping your developers write regex without weeping.

I do not want to be in charge of your company's life-or-death decisions while your data pipes are held together by digital duct tape and prayer.

When the very people training the frontier models start waving the caution flag and admitting that we are outpacing our own ability to understand internal model states, the debate is officially over. Pacing is not surrender. Pacing is basic engineering hygiene.

Here is the blunt truth for every business leader currently staring at their screen in total confusion:

• You don't need an autonomous AI swarm to approve a standard purchase order. A boring, deterministic database workflow will do that without trying to rewrite the rules of physics.

• Messy data plus cutting-edge AI just equals automated chaos at enterprise scale. Clean up your foundation first.

• The human in the loop isn't a bottleneck. The human in the loop is the only reason the system doesn't drive off a cliff when a model hallucinates a new tax bracket.

Welcome to the transition era. Put down the hyperventilation paper bag, stop feeling guilty for not turning your entire company over to an algorithm, and let's get back to building real systems with real human oversight.

Now, if you will excuse me, I am going to enjoy having a governor on my engine so I don't accidentally try to optimize the global internet into paperclips.