The current state of technology is AI - or large language models, to be precise. Everyone is excited about it. It will change our lives and disrupt entire markets. They say. Society is no longer driven by facts, but instead narratives that fits into whatever political agenda. That being to please shareholders, or fear of missing out on something. It is good to be excited about something, this all comes at a cost.

Not everyone is excited about AI. In the US, a study revealed half its participants are nervous when it comes to AI. The amount of people engaged being enthusiastic is falling. I personally recognize myself in this development where my enthusiasm for AI spiked within the first few months ChatGPT became a thing. After this, it became a tool much like every other piece of technology. My life does not revolve around AI - it is an utility that makes some of my work easier.

Compared with many other technological shifts, it is not technical people driving the AI shift. It is management, shareholders, and LinkedIn. “AI first” is a strategy almost every corporate workplace has heard of. The strategy is always the same: Find a use-case for AI, some cases mandatory trainings on how to use AI. Much of the motivation is to drive AI adoption and report that the business is “AI enabled”. This shows that the company is ahead of the curve. This has gone to the point where a copypasta on AI-enablement is circulating on reddit.

Not everything is doom and gloom with AI. Knowing when to use it and understanding its limitations is key to know how to actually make value with AI. This is when it has already been trained on a similar use-case it is being prompted at, which will help out in some cases. Asking on topics it has little training data on typically results in hallucinations, meaning you will not find an answer to your question in Google. There are several challenges that remains unsolved with ho we adopt AI:

• The AI responses can be very convincing - how will we avoid its responses making us delusional from overestimating our own abillities?
• How will we maintain ownership with the generated output? Who is responsible when something breaks or is wrong?
• How will we drive innovation and create new programming languages, tools, and frameworks when we constrain ourselves to generating based on previous knowledge? How will juniors learn and develop into seniors?
• What is even the point of having a programming language when we constraint ourselves to vibe coding a project and disregard tge output? Many people tend to forget we have had technologies enabling people without technical skills to create webpages since long ago. Remember Adobe Dreamweaver and Microsoft Frontpage?

Currently, way too few people speak up about these problems and ask critical questions. As a consequence of the top-down decision of “AI first”, many are too concerned that they might be replaced if being too negative. Many people I work with have become surprised at how frequent I use AI tools despite asking critical questions regaring how we adopt the technology. Even though management pushes for AI-first, it is also part of our responsibility as technologists to ask questions back in order to drive a sustainable and responsible adoption of it.

As a consequence of how the “AI shift” is being pushed, many people are talking about the AI bubble. Circular investments between model vendors and hardware manufactures, and uncertainty when AI will deliver return on investment. There are many narratives circulating around this from that the market will crash because white collar jobs will become automated thanks to AI, that the market will crash due to the circular investments popping as for example OpenAI struggles to find a sustainable business model, or data center hardware supply chain halting because of the middle east war leading to a vacuum for the available AI compute.

Today the focus is shifting towards on the costs of AI. AI compute has been heavily subsidized and now the costs are starting to rise. OpenAI is shutting down Sora to “focus more on core products”, Anthropic bans OpenClaw after it started burning up available AI compute, and GitHub Copilot is cranking up their model cost multipliers. Now Microsoft and Meta is laying off people to justify their AI investments. Several companies have started restricting the availability of AI compute to their employees due to costs. Other companies attempts to put lipstick on the pig by telling how innovative they are with how many tokens they use. This lead to token maxxing become a trending new term.

There is no doubt that AI is here to stay. The way AI is currently being pushed has too many similarities with other bubbles such as the dotcom bubble, and the web is still here to this day. In ten years many of the places we see AI could be gone. How all of this spans out depends on how society adops AI and which usecases we will find for it. But in the end, was it really worth it? Which narrative do you chose?

OpenClaw (previously called MoltBot, and ClawdBot) is the latest on the AI hype train. In short, it is an integration platform that connects whatever digital services that the bot owner has towards an LLM. This opens up for creating a digital assistant that gets access to email, calendar, and whatever services it is hooked up to. The LLM can be seen as the brain, while OpenClaw itself contains the state of the bot. OpenClaw quickly became the most starred repo on GitHub and is argued to be one of the reasons why self-hosted LLMs are ranking high on sites monitoring LLM traffic. Recently NVidia lauched NemoClaw which is a wrapper arround OpenClaw.

Why would you want a Mac Mini for OpenClaw?

Countless of videos and blog posts about OpenClaw follows the pattern that they open the video with “that you don’t need a Mac Mini to run OpenClaw”, with the remainder of the video how you would set it up. To clear things up, these are the reasons why someone would want to actually use a MacMini for this (they would typically never mention this in the blogs or videos):

• The bot runs on its own device, which some find appealing as it runs on its own little physical box. This way you can put a sticker “HomeBot” on it.
• People argue the unified memory with MacMini is good for running local LLMs (such as Qwen), but many argue these are not sufficient for OpenClaw and end up using external LLMs instead.

To me this is frustrating. Not only because I bought a Mac Mini (sadly with just 16 GB of ram), but because it gives me the vibes that someone is making me think I would need a Mac Mini for running this locally and this is not the case. Adding onto how so many of these posts look similar, it does not make the posts trustworthy and it appears more as a scheme.

Security by Optimism and Prayer

Having a system integrated into personal data opens up the risk of having this being leaked out considering the LLM itself is the engine making decisions on basis of which data the user has access to. OpenClaw has some security mechanisms where the bot owner can define some guardrails on what it should not do. For example “Don’t exfiltrate private data. Ever.”. These are instructions fed into the LLM to prevent it from misbehaving.

This security architecture can be compared with the following: Imagine a bank where customers are only asked to “Don’t withdraw more money than you have, and do not rob the bank. Ever.” without having any system that limits customers from robbing the bank by design. The bot owner has to trust that the LLM does not misbehave instead of knowing that there is a deterministic system that will not decide to empty someones bank account over night.

It is possible to restrict OpenClaw from integrating with certain systems by making it ask before making actions. However, this renders the entire purpose of OpenClaw rather pointless as it should be able to act on its own.

Fundamental Security Flaws

There are mainly two issues that prevented me from starting to use OpenClaw:

1. Having a model that runs locally which is actually good. Sending the amount of information OpenClaw would need to a big tech company is not something I see as a viable option.
2. Knowing that the bot will not start misbehaving, either because it decides to, or because someone is gaslighting it to do that. How can it be secured against prompt injects? How can it be secured against hallucinating and deciding to make destructive actions? How can it be secured against manipulation from other actors to leak data?

The first issue is easier to solve than the second, but it could also impact the second. The second is an architectural constraint of how an LLM works. This issue can be compared with how OpenAI’s agentic browser Atlas is vulnerable to prompt injections. On top of this, the bot will act as a remote code execution engine which will pick up content and decide on its own whether or not to make actions on what it sees on the net.

Seeing users of OpenClaw get surprised when their bank account is emptied, or their email account is purged reminds me of when people discovered that people could lie on the internet. My journey with OpenClaw ended with the two major issues I mentioned above before I got to set it up. And I do not find this reasoning very controversial. I love the use-case of it but unless these isseus are adressed, it is not something I want would use - and I would go as far as advicing others from using it. Seeing people set up this without much thought into the risk (and then have it backfire) appears as irresponsible use of technology. While some argue we should meet AI with “openness, curiosity, and willingness to learn”, it appears that critical thinking was left out of the process. As with most things related to generative AI, OpenClaw looks good on the surface, but when you dig into it, things fall apart.