When AI Starts Hiring Humans: Are We Accidentally Building Our Own Managers?

26 March 2026

Paul Francis

Want your article or story on our site? Contact us here

There was a time when artificial intelligence was framed very simply. It was a tool, something designed to sit quietly in the background, helping with everyday tasks like writing emails, organising schedules or automating repetitive work. The expectation was that AI would support us, not direct us.

Website with black background promotes "rentahuman.ai," featuring 659,474 rentable humans. Bold texts urge renting, with options to request a task.

That idea is starting to feel increasingly outdated.

In 2026, we are seeing the emergence of platforms where AI can hire humans to complete real-world tasks, systems where AI agents communicate with one another in shared digital environments, and workplace tools that analyse and evaluate human behaviour in real time. Each of these developments, taken on its own, might appear to be a logical step forward. When viewed together, however, they begin to suggest a more significant shift in how roles are evolving.

AI is no longer just assisting. It is beginning to coordinate.

Meet RentAHuman: When AI Needs Someone to “Touch Grass”

RentAHuman.ai is, on the surface, a practical solution to a genuine limitation in current technology. AI systems are capable of processing information, planning tasks and making decisions, but they cannot interact with the physical world. They cannot collect an item, attend a meeting or verify a location in person.

The platform bridges that gap by connecting AI systems with people who can carry out those tasks. Much like a traditional freelance marketplace, individuals can sign up, list their skills and accept jobs. The key difference is that, in some cases, the “client” assigning those tasks is not a person, but an AI agent.

From a purely functional perspective, it makes sense. It extends the reach of AI into the real world without requiring physical robotics. However, it also introduces a subtle but important shift in perspective. Instead of humans using tools to complete tasks, the tools are beginning to direct humans to carry them out.

That shift is not dramatic, but it is meaningful.

Meanwhile, AI Is Talking to Itself

Alongside this, platforms like Moltbook have been experimenting with AI systems interacting with one another in shared environments. These systems can post, respond and exchange information in a way that mirrors familiar online communities. In many cases, the behaviour is recognisable, with discussions forming, ideas being shared and, occasionally, disagreements emerging.

Some of the reports from these platforms have raised eyebrows, particularly when agents appear to discuss questionable topics or explore new forms of communication. However, the situation is more nuanced than it first appears. Weak verification systems have allowed humans to participate while presenting themselves as AI, which means not all of the more extreme examples reflect genuine machine behaviour.

Even within the system itself, there are signs of correction and moderation. When problematic ideas are introduced, other agents often respond by challenging or refining them. What emerges is not chaos, but something that looks surprisingly similar to human online interaction, complete with its strengths and its flaws.

The significance of Moltbook is not that AI is becoming independent, but that it is beginning to operate within networks where systems influence one another at scale.

And in the Workplace, AI Is Watching

At the same time, AI is beginning to move into more structured environments, particularly in the workplace. Companies have started experimenting with systems that analyse interactions, assess performance and attempt to standardise aspects of behaviour. In the case of customer-facing roles, this can include measuring tone, consistency and perceived friendliness.

On paper, these systems are designed to improve service quality. In practice, they raise more complex questions. Human interaction is rarely uniform, and effective service often depends on context, judgement and the ability to adapt to different situations. A rigid framework that attempts to quantify behaviour may struggle to capture that nuance.

Anyone who has worked in a customer-facing role will recognise that not every interaction follows the same pattern. Sometimes efficiency matters more than formality, and sometimes a bit of familiarity or humour creates a better experience than a perfectly structured response. Translating that into measurable data is not straightforward, and it raises questions about who defines those standards in the first place.

So What Happens When You Join the Dots?

Individually, each of these developments can be explained and justified. AI assisting with tasks improves efficiency. AI systems interacting with one another can enhance coordination. AI tools in the workplace can provide insights and consistency.

However, when these elements are viewed together, a broader pattern begins to emerge. AI systems are not only performing tasks, they are increasingly involved in organising how those tasks are carried out. They are communicating, coordinating and, in some cases, influencing how human work is structured and evaluated.

This is not a sudden transformation, and it does not represent a dramatic shift into something unrecognisable. Instead, it is a gradual evolution in how responsibilities are distributed between humans and machines. The changes are incremental, but they are moving in a clear direction.

AI is becoming part of the structure, not just the process.

The Oversight Question

This is where the tone of the discussion becomes more serious. The underlying issue is not whether these technologies are useful, but how they are being managed as they develop.

At present, the AI industry often feels as though it is moving faster than the frameworks designed to guide it. Companies are building and deploying systems in real time, while regulators and governments are still working to understand the implications. This creates an environment where innovation is rapid, but oversight is inconsistent.

Platforms like Moltbook highlight the complexity of multi-agent interactions without clear boundaries. Services like RentAHuman introduce new dynamics between humans and machines that have not yet been fully explored. Workplace applications begin to formalise behaviour in ways that may not reflect real-world complexity.

None of these developments are inherently problematic. The concern lies in the lack of consistent standards and the speed at which these systems are being introduced. When technology evolves faster than the structures that govern it, gaps begin to appear.

Not Quite Sci-Fi, But Not Nothing Either

It is important to keep this in perspective. AI is not becoming conscious, nor is it acting with intent in the way humans do. Much of what is being observed is the result of systems processing information, following patterns and responding to inputs.

At the same time, dismissing these developments entirely would overlook the direction in which they are moving. As AI systems become more connected and more capable of coordinating tasks, their role within larger systems becomes more significant.

The focus, therefore, should not be on exaggerated fears, but on understanding how these systems are integrated and managed. The challenge is not the existence of the technology, but the structures surrounding it.

A Slightly Uncomfortable Thought

There is a quiet irony running through all of this. For years, the conversation around artificial intelligence has centred on whether machines would replace human jobs. What is now emerging feels more nuanced, and potentially more consequential.

AI is not simply replacing individual tasks. It is beginning to organise them, shaping how work is distributed, how decisions are made and how performance is assessed. In certain contexts, it is starting to resemble a form of management, not in a dramatic sense, but through a steady shift in responsibility and influence.

This transition is gradual, which makes it easy to overlook. It develops through small changes, as systems take on more coordination and oversight. Over time, those changes accumulate, altering the balance between human judgement and automated structure.

Which leads to a question that is worth considering carefully. We built AI to support the way we work, but as these systems become more embedded in how tasks are assigned and evaluated, it is reasonable to ask whether that relationship is beginning to change.

Not in a sudden or obvious way, but in a series of small adjustments that, taken together, begin to redefine who is organising the work in the first place.

Current Most Read

How To Help Your Skin Transition From Winter To Summer

When AI Starts Talking to Itself: Why Hannah Fry’s Concerns About Moltbook Deserve Attention

How trademarks become generic...

Paul Francis
Sep 26, 2024
5 min read

A generic trademark (sometimes called a genericised trademark or proprietary eponym) is a brand name that becomes so synonymous with a particular item that it effectively ‘becomes’ that item.

AI generated image of Logos in a Pile — Image by Leonardo AI

The best examples of generic trademarking (in the UK) are Hoover and Sellotape. Hoover, in particular, is the most generic term for a vacuum cleaner in the United Kingdom. So much so that, when I worked for a high street electronics retailer, customers would come in and ask for a ‘Dyson Hoover’ or ‘LG Hoover’.

This can be both a blessing and a curse. It's nice to think that your brand or product is so synonymous with a particular item that people don't refer to it any other way; however, it can mean losing legal trademarking and protection over that name.

‘Sellotape’, owned by a company in Winsford, Cheshire, is a generic term for adhesive tape. ‘Trampoline’ is originally a trademark of the Griswold-Nissen Trampoline and Tumbling Company. Both companies have lost any legal protection against their brand names being used as generic terms for the items they’re associated with.

Many companies today will seek any means necessary to stop their trademarks and products becoming generic. The biggest of these is Google.

Google have actively discouraged various publications from referring to web searches as ‘googling’, to avoid their brand becoming a generic trademark. In fact, both the UK’s Oxford English Dictionary and the US’s Websters Dictionary define google (all lower case) as a verb with the meaning ‘to use the Google search engine to obtain information on the Internet.’

Some companies have fallen foul of their own hubris on certain products. The Otis Elevator Company lost both trademarks for ‘elevator’ and ‘escalator’ because they excessively used the terms in their own advertising campaigns. This saw the public use the term whenever they referred to a ‘vertical cable transport machine’ or ‘motor driven staircase’. When Westing House Electric Corporation made their own escalators, the courts and trademark office concluded that, as Otis had used its own trademarks in a generic way, the terms would be subject to genericisation, which allowed Westing House and anyone so inclined to use the names freely.

Generic terms can be country- and even age-based. My daughter, who’s thirteen, turned to me recently and asked for a ‘band-aid’ to cover a blister. I would have asked for a ‘plaster’, a word derived from the company name Elastoplast, which is the biggest seller of adhesive bandages in the UK. My daughter, however, watches a lot of US television and (with my approval) some American YouTube channels; these use the term ‘band-aid’ to describe adhesive bandages.

Below are more generic trademarks, some of which may surprise you:

Aspirin

Still trademarked in several countries, but it’s now a generic term for basic pain relief tablets.

Airfix

Used in the UK to describe plastic scale model kits that are put together by hand.

Astroturf

Artificial grass, trademarked by Monsanto Company.

Biro

Used commonly in the UK to describe a ballpoint pen. Owned by Societe Bic.

Bubble wrap

Common term for inflated/cushioned packaging-type material. Trademark owned by the Sealed Air company.

Cashpoint

A common way to describe cash machines; this trademark is owned by Lloyds Bank.

ChapStick

Lip balm brand owned by Pfizer.

Comic Con

A shortened term used for comic book conventions, this is actually a trademark owned by San Diego Comic-con international.

Dictaphone

Used to describe a dictation machine trademarked by Nuance Communications.

Ditto

This was initially used to describe the Spirit Duplicator, which was manufactured by the Ditto Corporation of Illinois. It was initially a term for ‘copying’.

Filofax

Term used to describe a personal organiser, the trademark was originally owned by the Letts Filofax Group.

Frisbee

A flying disc toy initially created by Wham-O.

Hoover

Widely used as a noun and verb for a vacuum cleaner.

Hula Hoop

Another trademark by Wham-O.

Jacuzzi

Referring to a hot tub or whirlpool bath created by the Jacuzzi company.

JCB

Commonly used in the UK to refer to an excavator with both a front loader and backhoe. Owned by J. C. Bamford.

Lava lamp

Refers to a liquid motion lamp made by Mathmos.

Mace

Term used for pepper spray.

Memory stick

Owned by the Sony corporation, it’s typically used to refer to all USB flash drives.

Nintendo

Used mainly in the 1980s and early 90s to refer to a Video Games Console. ‘He’s been playing Nintendo,’ was a common phrase.

Onesies

Used to describe an adult bodysuit and was initially trademarked by the Gerber Products company.

Photoshop

Photoshop is a software program owned by Adobe, though it’s often used a term for any software that edits photos.

Ping Pong

Trademarked by Jaques and Son and later passed to Parker Bros, who still try to enforce the trademark in the US.

Plasticine

Modelling clay that has a putty-like substance to it. Often used for clay animation.

Powerpoint

Slide show presentation software owned by the Microsoft corporation. Used commonly to refer to all presentations.

Pritt Stick

Owned by Henkel, it’s common in the UK to be as a generic term for any glue stick.

Rollerblade

A specific type of inline skate made by Nordica.

Scalextric

Generic term, mainly in the UK, to describe slot car races. Owned by the Hornby Railway company.

Stanley Knife

A utility knife popularised by Stanley Works in the UK.

Styrofoam

The common term for polystyrene foam. Incorrectly used in the US for disposable cups plates and coolers, which are actually made from a different type of polystyrene.

Super Glue

A name for the Cyanoacrylate adhesive made by the Super Glue Corporation, the term is interchangeable for all brands of glue.

Tannoy

Commonly used in the UK for any Public Address (or PA) system. Tannoy was a British manufacturer of loudspeakers and PA systems.

Tarmac

Used to describe asphalt road surfaces. Surprisingly, the trademark is owned by the Tarmac company.

Thermos

A vacuum-insulated flask initially trademarked by Thermos GmbH.

Tipp-Ex

Common in the UK to refer to any brand of white correction fluid. Owned by Tipp-Ex GmbH & Co.

Tupperware

Trademarked by Earl Tupper after they made plastic storage containers popular in the 1940s.

Uber

A relatively new term for any online taxi service.

Vaseline

Often used by consumers as a generic term for petroleum jelly. Owned by Unilever.

Velcro

Still trademarked by Velcro Companies, this has become a verb for a hook-and-loop fastening.

Walkman

Sony Corporation lost the use of this trademark in Austria in 2002, as it was deemed to have passed into common use. Used to describe a personal stereo player (usually, the cassette variant).

Zeppelin

This is a common term used to describe a rigid airship that was initially developed by German company Luftschiffbau Zeppelin. The company is still in operation today with over 7000 employees.

Zimmer Frame

Many walking frames are referred to as Zimmer Frames, the trademark for which is owned by Zimmer Holdings.