Operation ChatGPT – a fair go too far (Part 2)

By Alicia Lucas

Business Models

Uber adjusts its business model depending on the amount of resistance it finds in a new market but only as much as it has to, writes Dr James Muldoon, University of Essex, UK, and Ping Sun, now Associate Professor, University of Chinese Academy of Social Sciences, China and the Chinese Academy of Social Sciences, China. In unregulated markets, Uber implements the most flexible and profitable business model it can. Where states or cities have, or quickly enact, regulations to protect existing taxi companies and workers the company adapts in the most minimal way and sometimes this has meant it leaves the country. Driver or union protests can also be effective against some of Uber’s practices, however, Dr. Muldoon and Associate Professor Sun found unions sometimes made a deal with Uber that did not help Uber drivers but did benefit unions through improved membership.

Countries that didn’t just largely accept Uber’s terms and conditions have tended to preserve their own standard of living and/or independence in the taxi sector to some degree at least. Uber appears to keep finding ways to challenge regulations or better its position in most countries and finding chinks in the armour to re-entered countries they have exited. Uber left South Korea, Spain, and Denmark (amongst others) not long after starting in these countries. Uber returned to South Korea in 2021 by working with the country’s second largest conglomerate SK Group, producer of computer chips amongst other things. Uber re-entered Barcelona, Spain, during the Coronovirus pandemic, in part, because there were many unemployed desperate for work and despite protests by the local taxi industry. Uber returned to Denmark in 2025 by buying local taxi companies, including the largest Danish taxi service. The Danish Competition and Consumer Authority is now considering whether the latest acquisition will significantly reduce competition.  

China and Germany are more complicated and show how Uber can keep a toe in the water if it thinks warranted. Uber’s unsuccessful attempt to enter China saw it sell its Chinese business to a Chinese competitor, albeit retaining a small stake in that business. Uber is now partnering with Chinese company Momenta, a company producing cars with reduced reliance on human drivers/operators, to test these cars with the Uber app in Munich. Uber faced multiple challenges in Germany that limited its advance. Germany has also kept abreast with changes to technology. For example, Momenta has already collaborated with Germany’s Mercedes-Benz to include Momenta’s technology in cars sold in China.

In contrast to the countries that largely tried to prevent a then unprofitable company with seemingly unlimited funds from undermining their own industries, Australia has ended up with Uber holding a “market dominant position” or “monopoly position” according to Choice in 2025. Choice originally thought Uber would be good for competition but is now questioning this view. Uber overtook taxis use in 2018/19 and their lead has continued to grow according to Roy Morgan in 2025. Their survey indicates over three million Australians are using Uber in preference to taxis.

Australia, itself, has been disadvantaged. State payroll tax has declined and a tax haven was used to reduce Australian government tax. Uber no longer uses that tax haven but it possibly still pays insufficient tax. Taxi companies still have to pay taxes. Australian governments have recently been acting to reduce tax avoidance. Amongst other measures, company tax information by large multinationals will now be made public as the Australian Government passed country-by-country reporting rules in 2024 to provide a check by many eyes on such things as tax havens. Company tax information such as income and indicators of transferring of profits from one part of the organisation to another to avoid tax for each country in which it operates will be viewable by the public. Uber Australia paid about $81 million payroll tax on a $57 million taxable income while its total income was $3.5 billion in 2023-24. Sandy Spanos for Victorian Taxi Families in 2015 claimed the industry provided state and federal governments with about $1.4 billion in tax and GST. Spanos also said the taxi industry provided 150 000 full-time jobs across Australia.  

Drivers for Uber need to pay tax and consider GST if income is sufficient. The tax collected from drivers prior to the arrival of Uber may having been greater than now. Uber is part of the “gig economy” in which drivers do not typically work full-time and it is only within the last couple of years that Uber drivers became covered for minimum wages with the Closing Loopholes amendments to the Australian Fair Work Act. Australian drivers working in “gig economy” conditions were found on average to be earning less than half the minimum wage in 6 Australian cities in a union-backed Australian Institute, Centre for Future Work study.

Like Uber, ChatGPT became available in Australia in a way less likely to directly overly antagonise anybody. OpenAI released ChatGPT for free. People were able to use it for seemingly inconsequential things such as fun, answering a question and curiosity. People began to see and explore the model to help study or undertake work. ChatGPT was able to amass support with only the appearance of vague concern from existing services in Australia. Sometimes media personalities  even expressed their lack of concern about the impact of machine learning (portrayed as “artificial intelligence”) on their types of jobs. It was different in the USA where actors and screenwriters went on strike to protect their professions in 2023 following its release in 2022.  

Both OpenAI and Uber appear to work on a divide and conquer strategy, particularly by involving vulnerable individuals rather than the “sharing” story Uber was considered a part of. For example, Uber garnered support from those who could leave behind unemployment by becoming Uber drivers and, hence, became pitted against existing taxi drivers. Likewise, those who use ChatGPT to help them deal with mental health issues rather than using a trained psychologist. Uber benefits from enticing the support of groups of people not just from lower prices. Users of Uber who can be motivated to protest in the company’s favour are also used.

In their everyday business operations some suggest Uber may be considered a “rentier”. Instead of renting out, for instance, houses, rent is sought for, say, fine-tuning advertising using data collected from people using the company’s digital app. Uber’s booking app connects passengers to drivers supplying an existing product or service – commercial passenger services – and this leads to a multi-sided medium of exchange explain Professor Kean Birch and D. T. Cochrane from York University, Canada. Rent can be based on data collected from all users of the app, being multi-sided. An additional rent is gained by Uber by taking a portion of the payment for each passenger ride.

Professor Birch and Cochrane regard digital rentiers as different from traditional rentiers largely because they are digital. Assets, like data, can be created digitally and there is an increased ability to gain rent such as from acting as a digital go-between rather than, for example, just collecting rent from one or more individuals for using a piece of land. Uber customers are effectively not only the passengers but also the drivers. While companies have always tried to understand customers to better target and sell products, where helpful, the amount and type of customer data collected and the ease in which it can be moved around seems to have given this data greater weight. This is believed to be how a company may make more profit but it is not innovative. It is, at most, an extension of existing practices.

Professor Birch and Cochrane argue Uber (and similarly Amazon) “inserted” digital services or products where companies already provided services or products. The expectation, the authors assert, is that these companies will eventually become monopolies and therefore will be able to extract monopoly rents. In the end, prices will increase for the purchasers and advertisers will pay more due to lack of competition.

The longer term possibility of a monopoly encourages investors notes Assistant Professor Petter Tornberg, University of Amsterdam, The Netherlands. Further, the Assistant Professor contemplates, research into potential problems of rentier companies is perhaps actually helping to fuel investment. The companies may be rentiers, however, he believes this is not the reason for their success – they appear to be subject to a non-conventional form of investment that helps pave their way and rentier research is “effectively advertising” Uber may become a monopoly.

Investment in these types of companies is motivated by its “likelihood to become subject to speculation” because of its monopoly possibilities rather than how much profit it will make the Assistant Professor argues. The story surrounding the product or service is, therefore, vital to investment, particularly, if there is a quick growth in users that makes monopoly rent seem more obtainable he explains. There may be a market bubble, but with a compelling story, investors ignore risk and anticipate profit “from their rapid inflation” and invest. Actual evidence of potential, such as increasing revenue, is of lesser interest, but, being able to extract data is critical. “Data” is described by some as the “new oil” the Assistant Professor notes. Data is part of the story.

These companies are distinguished by very large market valuations “even when running steep losses” the Assistant Professor finds from the literature. Data spurs investment for two main reasons. It is portrayed as the producer of previously unheard of efficiencies to a company although the Assistant Professor states there is limited evidence to support this contention. He observes firms extracting this data, including Uber, “are at the same time remarkably inefficient: even with the business advantages of dodging regulation and taxation, passing risks and costs onto workers, and imposing precarious and exploitative labour conditions” the companies “some how still struggle to make profits”. Uber started in 2009 and its first profit only occurred in 2023.

The second reason is the belief data can be used to predict and influence people’s behaviour but, here again, Assistant Professor Tornberg finds there is “little empirical evidence that targeted advertising actually works.” A return to old-fashioned mass-marketing may be worthwhile according to some researchers the Assistant Professor notes.

Ultimately Assistant Professor Tornberg posits the success of these companies relies on access to venture capital that allows them to survive and aggressively contest markets in the USA and overseas without needing to profit. Uber use venture capital to subsidise prices paid by passengers and drive user growth hence feeding the story that helps them gain further investment.  

When Uber did make a profit it has been put down to paying drivers less and charging passengers more. A study by Professor Len Sherman, Columbia Business School, New York, found Uber’s shift to “upfront pricing” (basically dynamic pricing) saw the company taking more than 42% of a trip fare in 2024 rather than the 32% of 2022. The study examined Uber trips in 8 major US cities between 2019 and 2025. Seattle’s fare for a 30 minute trip were the highest in the USA. Uber explained the high fare prices in Seattle resulted from local regulations including the impost of a minimum wage for workers. Professor Sherman said fare prices and Uber’s take were increasing independently of the minimum wage. In a 2023 Forbes article, Professor Sherman wrote Uber’s increase in revenue was driven by an “obscure accounting change”, trip fares increasing 4 times faster than the rate of inflation as well as increasing its take from drivers. He observed customer numbers and productivity had not improved.

A University of Oxford study likewise indicated Uber made more money following introduction of dynamic pricing in London, UK. Uber drivers, however, saw their pay packet decrease, pay become less predictable, inequality between drivers increase and spent more time waiting for jobs explained Associated Professor Reubin Binns, Jake Stein, Siddhartha Datta, Associated Professor Max Van Kleek and Professor Sir Nigel Shadbolt. Even though drivers were meant to be paid for time logged onto the app, including time waiting, as the UK Supreme Court had found they were employees, not contractors, Uber still only paid them when they were on the way to picking up and carrying a passenger this study found. What a driver earns sounds higher when the amount of time waiting for a passenger to come along is not included in the calculation. This study and the USA study did not report on the accuracy and usefulness of the new pricing systems.

Drivers – who are workers – who are people – can seem just a cog in an “outsourced” wheel Uber effectively adjusts to increase the company’s revenue. Drivers largely interact with the app – not to people working at Uber. Even when the app help, including a chatbot, can’t resolve a problem and human advice is needed, Uber drivers are directed to talk to a third party company community support representative according to Assistant Professor Emma McDaid (University College Dublin, Ireland), Professor Paul Andon (University of New South Wales, Australia) and Professor Clinton Free (University of Sydney, Australia). Serious situations where drivers have been “deactivated” – switched off just like a piece of equipment and unable to pick up passengers – are dealt with by the third party company.

Uber had appeared to have taken the word of the passengers over that of the driver. Perhaps this is because of the simplistic maxim “the customer [the person paying the bill] is always right” or perhaps Uber only saw the passenger as being consequential in the longer term as “driver-less” cars maybe the long term aim.

ChatGPT’s potential business model is much more involved than Uber but concentrating on ‘psychological help from chatbot companions’ may break it down to be approximately equivalent of Uber’s ‘commercial passenger drivers’. Like the Uber drivers app, OpenAI rolled out ChatGPT across the world without asking if it was wanted or with little apparent concern of possible negative consequences to the receiving community. Like Uber, OpenAI are information technology specialists. The company is not comprised of psychological or psychiatric experts. One of the functions of ChatGPT was to provide companionship.

ChatGPT was not alone in providing a ‘companion’. A number of companies have released ‘AI chatbot companions’. ChatGPT is amongst the most recognised. In general, machine learning chatbot companions are available for anyone who wants technological support provided in some way. It could be just to “talk to” but can even be to address mental health problems.

Concerns are being raised that these types of programs may lead to: unhealthy dependencies on technology; minors being exposed to explicit content; people committing crimes; or even suicide Samual Greengard advises. Professor Emeritis Allen Frances, Duke University, USA, and Luciana Ramos, student at John Hopkins University, USA, have questioned why these tools have been rolled out with little consideration of the consequences for people’s mental health, for example, no mental health professionals were involved in model training.    

This is not a trivial problem. Parents of a California, USA, teenage boy who committed suicide are suing OpenAI in the San Francisco Supreme Court the Geelong Advertiser reports using an article originally published by the New York Post in 2025. ChatGPT, amongst other things, allegedly offered to produce a suicide note and advised on the type of knots suitable for hanging. A criminal case referred to by Greengard was that of a 21 year old who broke into Windsor Castle to kill the Queen of Great Britain (and Australia) following using chatbot Replika as reported by the BBC in 2021.

On anthropomorphic harm, the authors raise the concern about not only having machine learning respond in sentences that look or sound human but that are now presented with human-like emotion. They note that chatbots  “are remarkably good at mimicking it [emotion] and inducing it in humans” and “… how easy it is for humans to fall in love with seductive bots”.

In a followup article, Professor Emeritis Frances hopeful expectation that OpenAI would successfully tackle mental health problems after the company admitted ChatGPT causes psychiatric harm was tempered by past actions. The not-for-profit OpenAI’s original altruistic aims including protecting humanity “from the potential harms of artificial intelligence” were replaced by an emphasis on profit once competitor Google’s machine learning developments showed real promise. Business integration was enabled by encasing ChatGPT in a for-profit subsidiary and gaining private investors in 2019, including Microsoft and others from Silicon Valley, who would want a return on their investment. Like Uber, investment is huge and profits are not expected for a number of years.

The first indication OpenAI may be considering moving into mental health services using ChatGPT in a way similar to how Uber conducts its business comes from descriptions of ChatGPT’s future on the OpenAI website following the raising of mental health issues. OpenAI is looking at developing a “network of licensed professionals” that could be contacted through ChatGPT, that is, be the go-between between the suppliers and the buyers. There are other steps they plan to do first such as connect people to resources in the USA and Europe followed by “other global markets” as well as detecting problems before the user has moved into acute crisis and put them in touch with certified therapists. This seems to be a similar process as Uber used where they first rolled out the more conforming UberBlack before UberX.

To guide progress and perhaps indicating “well-being” as an UberBlack equivalent, OpenAI has now created an “Expert Council on Well-being” to advise on healthy interactions with machine learning and will “work” with relevant experts from the Global Physician Network.

It is interesting to ponder why the response to concerns regarding mental health released by OpenAI in August 2025, did not mention, as observed by others, any need to address the possible negative effects of imitation of human emotions. The company appears to be doubling down and says once a problem is detected, the model will shift to using “supportive empathic language” and “steer” the ChatGPT user to help. Like Uber, OpenAI may only respond to concerns as much as it feels necessary to realise its plans.

The second indication an “Uber-like” mental health service is possible comes from a study that examined copyright and data protection, amongst other considerations, and reached the conclusion it is possible companies are positioning generative machine learning models to play an intermediary role between a supplier and a buyer similar to that of companies like Uber. If the results of ChatGPT were to be used, say, in someones business, the business would not have to pay OpenAI for the results and the business would be responsible for acknowledgement of copyright and copyright infringement reported UK researchers Professor Lilian Edwards, Newcastle University, UK, Assistant Professor Igor Szpotakowski, University of Leeds, UK, Gabriele Cifrodelli and Josephine Sangare, both from University of Glasgow, UK, and James Stewart, University of Edinburgh, UK.

OpenAI appears similar to Uber in that users of ChatGPT bear the responsibility for legal action taken against them regarding copyright just as drivers were fined in the early days of the introduction of Uber.

Professor Emeritis Frances raised his concerns about the OpenAI CEO and concerns were also raised about Uber’s first CEO and the culture under his leadership. In its early days Uber appeared to think it could do whatever it takes to win although this is unlikely to be the level the OpenAI CEO would go to. Leaked documents, known as “The Uber Files” included evidence the company lobbied government officials to stop investigations, reduced tax by sending profits through tax havens and could immediately stop access to the companies servers if a government raid was imminent. Uber seemed to think it was okay to mislead or, perhaps, even under-report problems to protect its bottom line. Uber didn’t tell anyone for over a year personal details of 57 million customers had been hacked in 2016 and pretended it was part of a “bug bounty” program. The company payed the hackers, once tracked down, US $100 000 and forced their signing of non-disclosure agreements. Uber’s CEO was made to leave after this and other problems in 2017 although Travis Kalanik sat on the board until the end of 2019.

Reports largely suggest the replacement CEO, Dara Khosrowshahi, has turned the culture around at Uber but there are signs something is still not right. In 2022, Uber was fined $21 million after the ACCC took Uber to the Federal Court and was found to have mislead customers over a number of years about cancellation fee charges and for making the price of “Uber Taxi” rides, as distinct from other Uber rides, mostly higher than they actually were preventing consumers from making an “informed decision about their transport choices”. In 2024, Uber was fined 246 million pounds for transferring driver’s sensitive information to the USA over a two year period in violation of EU rules. Uber is contesting the fine. Uber is currently being investigated by a USA House committee for under reporting sexual assault and sexual misconduct. Emily Steel identified more than 3000 lawsuits against Uber from passengers alleging drivers sexually assaulted or harassed them.

To be continued…

 

Part 1:

Operation ChatGPT – a fair go too far (Part 1)

Keep Independent Journalism Alive – Support The AIMN

Dear Reader,

Since 2013, The Australian Independent Media Network has been a fearless voice for truth, giving public interest journalists a platform to hold power to account. From expert analysis on national and global events to uncovering issues that matter to you, we’re here because of your support.

Running an independent site isn’t cheap, and rising costs mean we need you now more than ever. Your donation – big or small – keeps our servers humming, our writers digging, and our stories free for all.

Join our community of truth-seekers. Donate via PayPal or credit card via the button below, or bank transfer [BSB: 062500; A/c no: 10495969] and help us keep shining a light.

With gratitude, The AIMN Team