More investors are using chatbots to research shares, funds and pensions. It is easy to understand why. AI is fast, inexpensive and capable of processing far more information than any individual adviser.
AI investment advice is not currently the same as regulated financial advice. But is that because the technology is incapable, or simply because it has not yet been given the same information, access and legal responsibility?
Why the FCA is concerned
The Financial Conduct Authority recently warned that investors may be placing too much trust in general AI tools.
Its research found that 56% of investors aged between 18 and 40 trusted AI to help them make financial decisions. Four in five less-experienced investors had used AI for investment help, while 38% believed it was acceptable to make a decision based entirely on an AI response.
The biggest misunderstanding concerns regulation. Some 44% wrongly believed AI-generated financial information was regulated, while around a third thought they might receive compensation if it caused them to lose money.
As explained in Why financial watchdog is warning Britons not to take AI investment advice, general chatbots such as ChatGPT and Gemini are not regulated financial advisers.
That does not mean AI cannot analyse investments. It means there may be no regulatory protection if its information is wrong.
What if AI is given all your financial details?
One of the main arguments in favour of a human adviser is that they understand the customer’s personal circumstances.
They will look at someone’s income, age, debts, savings, pensions, tax position, family commitments and attitude towards risk. That clearly gives them an advantage over a general question asked through a chatbot.
But what if all those details are also given to AI?
A sufficiently advanced system could ask the same questions as an adviser and build a detailed financial profile. If connected securely to bank accounts, pensions, investments and live market data, it could potentially know more about someone’s financial position than an adviser who sees them once or twice a year.
AI could also monitor those details continuously. It might spot rising expenditure, falling income, excessive charges or an unbalanced portfolio before the customer or adviser notices.
The real issue is not simply that AI lacks personal information. It is whether that information is accurate, whether AI interprets it correctly and who accepts responsibility for the recommendation.
There are also obvious privacy concerns. Giving an AI system access to someone’s entire financial life would require extremely strong security and clear rules governing how that information is stored and used.
What does an experienced adviser offer?
An experienced adviser brings judgement, accountability and knowledge gained from dealing with real people.
They may recognise when someone is overestimating their tolerance for risk or chasing an investment simply because it has recently performed well. They may also stop a frightened customer from selling everything during a temporary market fall.
A regulated adviser must explain why a recommendation is suitable and keep records supporting that decision. If the advice is unsuitable, the customer may be able to complain or seek compensation.
AI may eventually perform much of the analysis better and faster. What a general chatbot currently lacks is responsibility for the result.
Human advisers are not infallible either. They can make mistakes, favour familiar products, charge high fees or allow their own investment preferences to influence their recommendations.
This is not a comparison between an unreliable machine and a perfect human. Both can get things wrong, but only one is presently operating within a regulated advice service.
Could AI investment advice be manipulated?
This is another question that deserves more attention.
AI systems learn from and retrieve enormous amounts of information. What happens if some of that information has been deliberately distorted?
Companies, investment promoters and fraudsters could attempt to flood the internet with positive articles, fabricated reviews, misleading performance claims or apparently independent commentary. The aim would be to make a particular company, fund or investment appear more credible when AI systems research it.
We have already seen businesses trying to influence search-engine results. It is reasonable to assume that some will also try to influence the information used by AI.
There is also the question of who controls the system. Could a financial platform quietly favour its own products or those paying it the highest commission? Would the customer know whether a recommendation was genuinely independent or partly shaped by a commercial arrangement?
A fluent and confident AI answer could make this particularly dangerous. People may assume it has independently examined the entire market when it has actually been working from incomplete, biased or manipulated information.
Regulated investment AI will therefore need clear rules covering its data, commercial relationships and how recommendations are produced. Independent testing and proper records will be essential.
What if AI gives everyone the same advice?
There is also a wider risk to investment markets. Millions of people could ask similar AI systems the same questions and receive broadly the same recommendations.
If several leading systems favour the same companies, sectors or funds, huge amounts of money could be directed towards them. Rising prices might then appear to confirm the original recommendation, encouraging even more investors to follow it.
The same could happen in reverse. If widely used systems identify the same warning signs and recommend selling, large numbers of investors could react at once and make a market fall worse.
Financial markets have always suffered from herd behaviour, but AI could accelerate it. A small number of systems might eventually influence where millions of people place their money.
Future investment AI will need to consider market concentration rather than simply recommending whatever appears strongest from recent data.
Where AI already has an advantage
AI can compare charges, analyse asset allocation and explain financial terminology almost instantly. It can work through huge quantities of information without becoming tired or limiting itself to products it already knows.
It can also provide basic help to people who cannot afford traditional financial advice.
A properly designed system could monitor investments every day rather than waiting for an annual meeting. It could warn when a portfolio becomes too concentrated, charges increase or changes in tax rules affect the customer.
This could help close the advice gap by making financial guidance available to people with smaller pensions and savings.
How long before AI can do more?
Over the next few years, regulated firms are likely to introduce AI systems connected to verified financial information, live market data and customers’ records.
These systems could complete the fact-finding process, assess risk, compare products and prepare personalised recommendations. The FCA is already allowing companies to develop and test advanced financial AI through controlled programmes.
For straightforward decisions, such as selecting a diversified fund or reviewing pension charges, AI may be completing most of the work within two to five years.
The important change will be AI operating inside a regulated financial business. The company would remain responsible for the recommendation, even when the analysis was completed automatically.
Complicated decisions involving inheritance, tax, retirement or difficult family circumstances may continue to require human oversight. Even then, AI could complete most of the calculations and identify the available options.
Will AI replace investment advisers?
AI will probably replace parts of an adviser’s job before it replaces the entire role.
Research, calculations, product comparisons, report writing and portfolio monitoring can increasingly be automated. Advisers may manage more customers, while people with straightforward needs deal mainly with an AI system.
The likely result is a blended service. AI will conduct the analysis and continuous monitoring, while a qualified person reviews complicated cases and accepts responsibility for the final recommendation.
The question is no longer whether AI can produce a detailed investment plan. If it is given complete personal information and reliable financial data, it increasingly can.
The bigger questions are whether its information has been manipulated, whether its recommendations are genuinely independent, whether too many people are being directed towards the same investments and who is responsible when something goes wrong.
For now, general AI is best used as a research tool rather than an unregulated replacement for professional advice. As regulated firms begin giving AI access to verified financial information and customer data, its role will undoubtedly grow. This should be approached cautiously, with proper oversight, transparency and clear responsibility whenever AI influences an investment decision.
AI Tuition Hub covers these developments as part of our growing course library. Members receive access to more than 20 different financial courses, together with all our other AI training, for just £9.95 a month.