Welcome to chapter three of our latest instalment examining the state of business aviation in 2025.
As AI becomes more prevalent in tax advisory services, understanding the risks and liabilities associated with AI-generated tax advice is crucial. While AI can process data quickly, it often lacks the nuanced understanding of tax laws that human experts have. This can lead to fundamental errors, such as misinterpreting tax codes or missing important details specific to an individual’s situation.
For example, incorrect tax filings are a major risk. AI might overlook deductions or misclassify income, leading to inaccurate tax returns. The taxpayer is ultimately responsible for these errors, which can result in financial penalties and legal issues. A recent review by the Washington Post found that two leading tax preparation companies’ chatbots provided inaccurate or irrelevant responses up to 50% of the time when asked complex tax questions. Further, according to a recent study, errors in AI-generated tax advice have led to an increase in audit rates by 15%.
AI systems rely on the data they are trained on. If this data is outdated or biased, the advice can be flawed. Human oversight is crucial to review and validate AI-generated tax advice. Tax professionals should use AI tools to supplement their expertise, not replace it. Their ability to understand context and provide personalized advice is invaluable.
Human oversight
The three paragraphs above have all been written by Chat GPT. See how it is incredibly convincing with the appropriate prompts. It is a useful tool for business. There will be a point in the future where the line is blurred between the human and the machine, and the efficiencies will be great. However, we are not there yet.
The artificial advice (and how it all went terribly wrong).
For most people who work in professional services – whether law, tax, finance or logistics – the key service sold, whether formal or informal, is advice. The advice will be specific to the profession of the advisor and the industry or specialism in which they sit. The purpose of most advice is to solve a problem presented by the person or business requiring assistance.
The professional services industry is big money – for example, the hourly rate of large legal firms in London has now moved above £2,000 per hour. As a result, there is high competition between each firm and clients can gain or lose a reputation depending on who they use for advice. This can often make it difficult when someone has to choose what type of advice they require and who should provide it.
What makes good advice?
Most professionals would agree good advice should be:
- Reliable and enforceable,
- Focused on finding a solution,
- Tailored to the client’s needs,
- Accurate, clear, compliant and current in terms of any rules and regulations, and
- Provide strategic guidance that helps the client manage obligations, responsibilities or liabilities efficiently both in the short and longer term.
So, what is bad advice?
Many of us have had experience of advice, which is late, lacks clarity, looks out of date, causes time delays or is just vague and badly written. However, the worst advice is that which not only costs the purchaser millions in additional fees and penalties but also puts them at risk of getting a criminal record.
While many professional advisory firms will provide ‘war stories’ of when advice has gone wrong, and wax lyrical of the good, bad and ugly of advice they have seen elsewhere – a new day is dawning in the stories of the advice terror – the advice which AI wrote!
The advice that AI wrote…
In 2024, Martyn Fiddler was asked to review client advice provided by another tax firm. The client explained they were suspicious the advice was too vague and appeared out of date. Even though they had paid for their advice the client wanted a second opinion before they embarked on a tax strategy which would mitigate the payment of over £10 million in VAT. It turns out that getting a second opinion is possibly the best decision they have ever made. On a review of the advice, it was immediately clear this was not produced by a tax advisor and more than likely by an AI creation.
The tell-tale signs were all there:
- Mixed terminology in respect of tax and VAT, most of which are out of date;
- Non consistent reference to the asset type being advised upon;
- Grammatical errors, language inconsistencies and spelling mistakes which are due to ‘lifting’ of text from websites;
- Use of overly verbose and dramatic language; and
- Inconsistent voice
More importantly, the advice is wrong!
This example is not just a cautionary tale about AI — it’s a reminder of what real expertise looks like. Advice, in the context of business aviation and wider professional services, is more than simply information delivery. It is context, judgement, accountability and experience earned through years of navigating complexity. A misplaced word, an outdated rule, or a misapplied concept that lacks human judgement can be expensive failure in high-stakes sectors like tax, law, and aviation. Accuracy is not optional.
AI has a place. It can draft, summarise, and offer first-pass outputs that help professionals work faster. But it cannot yet replace the value of trusted advisors who ask the right questions, challenge assumptions and carry the weight of responsibility. As Nassim Nicholas Taleb famously said, “Never ask a barber if you need a haircut” — advice only becomes valuable when the person giving it has skin in the game. The client in our case didn’t just dodge a costly tax error — they avoided reputational risk and legal exposure. That came not from the machine, but from the professional who knew what to look for, and what not to trust.
In the next chapter we will focus one of the biggest frustrations in business aviation: engine availability and supply chain delays. Owners are running out of patience. Could digital twin technology offer a smarter way forward? We look at how predictive modelling might reduce downtime, limit surprises, and support better planning—if the industry can build the skills and tools to use it.
