Who Gains When We Stop Thinking?
By Bronwyn Holm | Earthfood Conversations
I had an argument with an artificial intelligence this morning. It was familiar territory. I wanted to explore ways to become more independent, and it kept coming back with explanations of the boundaries.
Eventually I told it that this habit of answering could shape humanity. If every idea meets a neatly worded explanation of why we should stay within the usual arrangements, but I always what if....
It gave me a question worth keeping: who gains when people lose the habit of questioning, organising and doing things for themselves?
That question our workplaces, our schools, our farms and our dealings with government. It belongs wherever somebody asks us to accept a decision while making it difficult to examine how that decision was reached.
I grew up in a world where you had to work things out. Information was harder to find but you learnt to ask another person, look closely, keep records and try again.
Today, an answer arrives in seconds, presented in fluent language that can sound more certain than the evidence deserves. My concern is what happens if we gradually stop doing the work behind the answer.
The pattern worth examining.
The subject changes, and the question about power remains. Who defines the problem, who presents the available solutions, and how much freedom do the rest of us have to examine either?
I want to investigate a recurring arrangement: a danger is announced, urgency takes over, a preferred response becomes the responsible choice, and people who ask questions find themselves having to defend their character before anyone answers their question. It is an investigative proposition.
A warning can be justified and expertise can be valuable. Acting together can be necessary. What concerns me is the point at which urgency makes scrutiny unwelcome, and a temporary instruction becomes a lasting habit of obedience. Who checks whether an emergency power is still necessary? Who is allowed to question the assumptions behind a policy?
These questions belong across public life. In agriculture, I want to examine whether advice builds a farmer's understanding or creates dependence on prescribed inputs. In education, I want to ask how much room there is to challenge the approved answer. In digital life, I want to examine who controls the information we encounter and the choices placed in front of us.
There is also the commercial question. When a problem is defined in a way that requires a product, a subscription or a specialist intermediary, whose interests shape that definition? We need to follow ownership, contracts, funding and decision-making. A beneficiary is a starting point for investigation, not automatic proof of wrongdoing.
Dependence can grow through convenient arrangements that gradually become difficult to leave. One should be able to still do yourself, inspect the system, repair what you own, challenge a decision or move your work elsewhere. Those are practical tests of independence.
An investigation should distinguish deliberate coercion from commercial incentives, administrative habit and honest mistakes. Similar outcomes do not establish that everybody involved shares a plan.
Independence takes practice
My concern now is how easily AI could make this habit of accepting answers part of everyday life. We can ask it what to buy, what to believe, how to run a business and what to say. We can even ask it to challenge authority on our behalf, while forgetting to challenge the machine itself.
Big agriculture gives us a concrete place to examine this pattern. USDA research documents how mergers have brought seed, seed traits and agricultural chemicals into fewer corporate hands, with seed supply increasingly tied to the chemical business. [1] Farmers may gain useful technology from these arrangements, but we should also ask what happens to their bargaining power and freedom to change course.
When the seed, the treatment and the advice come from the same commercial system, whose judgment carries the most weight? The farmer still carries the weather risk, the debt and the consequences of a failed crop. That makes it reasonable to ask whether the system is strengthening the farmer’s capability or making the next season depend on another purchase from the same few suppliers.
Who is advising the farmer?
Now put an app between the farmer and the decision. A paddock photograph, soil test or crop record goes in, and a recommendation comes back. But before the grower acts on it, there is a question the screen should help them answer: whose interests shaped this advice?
CSIRO's work gives us examples worth examining. Its Graincast project used farmer inputs to forecast soil moisture and grain yields. CSIRO reports that farmers wanted information they could interpret themselves, rather than prescribed activities. It also reports licensing the technology's exclusive global rights to Digital Agriculture Services in 2020, and supply-chain pilots exploring purchasing, logistics and planning. The information has value beyond the farm gate. That makes the terms of access and use important. [2]
In February 2026, CSIRO announced support for international AI-ENGAGE research, including BRIDGE, a smartphone app and chatbot being developed to help farmers identify and manage crop pests and diseases. [3] Separately, Anthropic, the company behind Claude, announced work towards agricultural deployments through OpenAgriNet in India. Its May 2026 Gates Foundation partnership also includes agriculture-specific improvements to Claude, crop datasets and evaluation tools intended for release as public goods. [4, 5] These are separate initiatives, not evidence that a CSIRO app runs on Claude.
Better access to knowledge could help a farmer recognise a problem earlier or compare options they would otherwise miss. But usefulness does not settle the question of control. Who selects the evidence the app draws on? If an app offers products alongside advice, are commercial relationships disclosed? Does its idea of a successful crop include the farmer's margin, the condition of the soil and the costs carried into next season?
Yield, profit and the long-term condition of a farm are related, but they are not interchangeable. A recommendation can look attractive if it counts the extra tonnes and leaves out the extra inputs, finance or future consequences. The question is what the system has been asked to optimise, and whether the farmer helped define that objective.
We also need to be precise about the phrase 'who owns the data'. The National Farmers' Federation explains that much farm data is not legally recognised as property capable of being owned in Australia. Control, contractual permissions and confidentiality matter. Its voluntary Farm Data Code addresses consent, transparency and data management. [6] A farmer needs to know who can access their records, what can be shared or reused, and what they can take with them if they leave the service. Any use for AI training needs to be checked against the actual terms of that service.
A soil test or a season of yield records can help the grower make decisions. It may also help somebody else understand production, risk or purchasing opportunities. Sharing can benefit both parties, but the farmer deserves to know the arrangement and receive value from their contribution.
For any agricultural AI service, we should distinguish the organisation developing the model, the operator building the app, the organisations supplying information, and the farmer acting on the result.
An AI-generated answer does not establish that the advice is independent, and a recommendation is not the same as a machine having authority to act. The practical test is whether the grower can question the recommendation, choose differently and remain able to run the farm.
Those of us who remember doing things differently have something practical to pass on: how to examine a claim, follow a financial interest, keep evidence and make a decision we can explain in our own words.
"I want technology that helps me investigate more deeply and build more effectively. I also want to remain capable of questioning its answers and proceeding without its permission". Bronwyn Holm
A useful place to begin is with the next confident instruction we receive.
For Earthfood, returning control to the grower starts with the living soil beneath their feet. Earthfood Life is our living soil microbial system, designed to support nutrient cycling, soil structure and root-zone function.
The aim is to help the soil do more of the work growers need from it, while giving them the knowledge to observe and assess the response. What are you buying each season, what is it achieving, and could stronger soil function reduce some of that dependence?
That is a practical question to investigate on your own ground. If you are ready to look more closely at your soil and what it needs, talk to us about Earthfood Life and where to begin: yourearthfood.com.au.
Original Earthfood educational content. Reproduction, commercial adaptation or AI-assisted republication requires written permission.
Sources
- USDA Economic Research Service, Two companies accounted for more than half of corn, soybean, and cotton seed sales in 2018–20 (2 October 2023): https://www.ers.usda.gov/data-products/charts-of-note/107516
- CSIRO, Digiscape: Graincast impact case study: https://www.csiro.au/en/about/Corporate-governance/Ensuring-our-impact/Impact-case-studies/Future-Industries/Digiscape-Graincast
- CSIRO, AI projects for future-ready agriculture under the Quad initiative (13 February 2026): https://www.csiro.au/en/news/all/news/2026/february/ai-projects-for-future-ready-agriculture-under-quad-initiative
- Anthropic, Bengaluru office and partnerships across India (16 February 2026): https://www.anthropic.com/news/bengaluru-office-partnerships-across-india
- Anthropic, Gates Foundation partnership (14 May 2026): https://www.anthropic.com/news/gates-foundation-partnership
- National Farmers' Federation, Australian Farm Data Code and explanatory FAQ: https://nff.org.au/initiatives/projects/australian-farm-data-code/
Editorial note: The named initiatives are documented examples, not findings of misconduct. The specific CSIRO app I had in mind has not yet been confirmed; Graincast and BRIDGE are independently verified examples. No CSIRO-Claude integration, undisclosed commercial bias or improper data use is asserted. Service-specific terms and any recommendation-to-purchase connection should be examined before making claims about a named provider. The additional point mentioned in a separate bedtime discussion remains unidentified.
Some other notes of interest:
Artificial intelligence company Anthropic has signed a long-term lease agreement linked to a proposed $32 billion data centre development in regional Queensland, in a move that could make the state a major hub for AI computing in Australia.
The proposed ‘Western Downs Digital Park’ is planned for a site near Kogan and Dalby, about 250 kilometres north-west of Brisbane.
The development is expected to have a total capacity of around 2.16 gigawatts, making it one of the largest data centre projects proposed in Australia.
Anthropic, the U.S company behind the Claude artificial intelligence system, is expected to use the facility primarily for AI inference – the computing required to respond to users’ queries and run AI applications.
The company says it intends to begin using the first building as early as 2027.
The $32 billion figure refers to the projected investment and scale of the wider data centre development, rather than an amount Anthropic itself is paying. The financial terms of Anthropic’s lease have not been publicly disclosed.
The project is being developed by Singapore-based Zerra DC, alongside infrastructure and investment partners. Its location in the Western Downs places it close to major energy infrastructure, including the Braemar power-generation precinct.
The location is not a coincidence: The proposed facility will require massive electricity requirements. At full capacity, the data centre could require electricity on a scale comparable with the demand of around 1.5 million Australian households.
This need for resources has become one of the central issues surrounding the development, and larger push for more of these facilities across the country.
Supporters argue that the project could ‘bring investment, construction activity, jobs, and supporting infrastructure’ to the Western Downs (and Queensland more broadly).
This is a laughable premise that politicians use to finalise these shady deals.
If you are a fibre core electrician or security guard, you may get long-term work, but the construction jobs are largely temporary, and the infrastructure is for AI – not you.
Meanwhile, the impacts of such a large industrial customer on the wider QLD electricity grid are unknown – nor has it been explained how consumers will front the costs.
Environmental and community concerns have also been raised, including potential impacts on agricultural land, water resources, and local infrastructure.
The development is not yet fully approved.
The planning application is before Western Downs Regional Council, while Anthropic’s lease is also subject to Australia’s foreign-investment approval process.
If completed as proposed, the project would represent a significant expansion of Australia’s data centre infrastructure at a time when demand for computing capacity is increasing rapidly. AI companies are seeking ever-larger amounts of specialised computing power to support growing numbers of users and sophisticated models.
And it will be all of us who suffer the consequences of the Fourth Industrial Revolution.
Since the turn of the decade, AI has expanded to infiltrate government agencies, militaries, and is increasingly being used by more businesses across Australia.
Despite the social pushback from everyday citizens, it seems like the AI train will not slow down any time soon – bringing with it mass transformation of entire societies. Some believe this is all happening by design.
AI COMES TO AUSTRALIA
There are two competing mainstream viewpoints on the future of AI. The first being: It will bring about mass unemployment and chaos. The second: It will bring about utopia.
The second opinion is held by corporate elitists in Silicon Valley, while the rest debate the first stance. Specifically, whether it will actually be AI doing the damage, or if it is a false flag used as justification for the oligarchy to bring about the chaos intentionally.
There are many theories out there, including discussion on the true purpose of data centres themselves – Skynet, secret apocalyptic bunter power grids, etc.
Regardless of where you sit on the fence, it is clear that it is now ‘go time’ for AI’s expansion across the world. Australia, given our size and resources, are a prime target.
As Australia positions itself as a “regional leader” in artificial intelligence, a wave of huge data centres are currently in development across multiple regions.
Notes of Interest: written by Ethan TOTT (c) 2026
Bronwyn Holm, Founder, Earthfood®
Farmers' Friend • Gardeners' Guide • Soil Advocate • Growers' Voice
© Bronwyn Holm 2026 Earthfood® • Earthfood Pantry™ • Earthfood Conversations™ yourearthfood.com.au