Cyber Territories - Dispatch #1
Signal #1.1 OpenAI’s e-commerce strategy has failed.
Source: https://futurism.com/artificial-intelligence/openai-pivot-into-shopping-disaster
Dispatch:
OpenAI is pulling the plug on 'Instant Checkout', a feature intended to revolutionise retail by enabling users to make purchases within the chatbot itself. Despite backing from giants such as Walmart and Shopify, users appeared virtually uninterested. It also proved difficult to adjust prices and products in real time, arrange refunds, and comply with local tax laws.
While many tech giants dream of becoming an omni-store or even the operating system of the internet, this seems to be a pipe dream for now. Amazon is best placed to solve the technical and logistical challenges, but lacks a successful chatbot. Google and Meta, on the other hand, struggle with a lack of consumer trust because their business model treats consumers as products. Therefore, what the Chinese company Tencent is achieving with WeChat is not easily replicable globally.
Reflections:
Is it wise to allow companies to grow into critical digital infrastructure when their business model treats people as products? If e-commerce does not work for tech giants, how will they recoup their substantial investments? What dangers lurk in their hidden despair? Is ultimate ease of use with everything on one platform more important to people than freedom of choice?
§§
Signal #1.2 AI language uses a limited set of rethorical figures compulsively. This results in a monotonous reading experience devoid of flavour.
Source:
https://www.deadlanguagesociety.com/p/rhetorical-analysis-ai
Synthetic image/AI-generated
Dispatch:
The em dashes that appear in almost all AI texts are annoying enough, but the compulsive use of rhetorical techniques such as antithesis is also starting to get on the reader's nerves. Once you start noticing it, you can't unsee it.
AI texts are full of expressions such as 'it is not this, but that'. AI slop is also increasingly peppered with uninspired tricolons combined with parallelism. This is not efficient, reliable or effective — it is simply annoying.
Reflection:
Why does AI language evoke such a visceral reaction from us? Despite their very broad training sets, how is it that LLMs remain limited to a few basic stylistic techniques in such an uninspired way?
§§
Signal #1.3 Google Workspace for AI agents: ‘You break it, you bought it’.
(Do no evil, Dear Customer)
Dispatch:
Google is making all Workspace apps, including Gmail, Drive, and Calendar, accessible to AI agents such as OpenClaw. Specifically, this update enables OpenClaw to access more than 40 different skills.
However, Google is passing on all the risks to companies that subscribe to their ecosystem. They are framing it as ‘not officially supporting’ it. In other words, they are saying: ‘We are not taking responsibility for this. We move fast, and when things break, you pay.'
Reflections:
How far can you encourage rapid innovation before it becomes reckless? How can you trust a company that does not dare to take responsibility for the tools it releases?
§§
Signal #1.4 The impact of AI on the labour market: more foam than beer
Source: https://www.anthropic.com/research/labor-market-impacts
Dispatch:
Anthropic has developed a new method for measuring the impact of AI on the labour market. Despite the theoretically high potential of AI for labour automation, this method shows that its actual application in the workplace is not really taking off.
Even in professions that Anthropic considers to have a high level of exposure to AI, such as programming and customer service, there has been no systematic increase in unemployment.
However, there has been a slowdown in the recruitment of younger employees, which may be due to management anticipating the potential effects of future automation.
Journalism appears to have moderate exposure, as some journalistic tasks are theoretically automatable. However, according to this method, there is no observable exposure. Craftsmanship is difficult to automate. You end up with more foam than beer.
Reflection:
Why don't we see any significant automation applications in the workplace, and why does it remain theoretical potential?
What is the ultimate social goal of achieving efficiency gains through AI automation? What will happen to knowledge transfer and the human learning curve if the current recruitment disruption causes a generation-sized gap?
§§
Signal #1.5 Intensive AI use can lead to 'AI brain fry' and reduced efficiency in the workplace.
Source: https://hbr.org/2026/03/when-using-ai-leads-to-brain-fry
Dispatch:
Professionals who use AI tools intensively every day are increasingly experiencing 'AI brain fry', a form of mental fatigue caused by excessive interaction with AI tools. This results in reduced efficiency due to increasing concentration problems and a higher risk of errors.
This phenomenon mainly occurs in situations where AI tools attempt to replace human creativity, reducing the user to a quality controller of the AI output. AI promises efficiency gains, less friction, more speed and greater volume.
This is achieved by centralising more tasks and responsibilities on a single human endpoint. However, the result risks being counterproductive and dulls the human capacity for creative and critical thinking.
Reflection:
How can we safeguard creative and critical thinking in the workplace?
What dangers arise from implementing AI in existing workflows without first fundamentally changing the operational design to prioritise people?
This blog is written by Patrick Lacroix in a personal capacity. AI tools are used for research, structuring, drafting and language support. All content is selected, verified, and edited by the author, who retains full editorial responsibility.