Follow any the development of any tech and you’ll find a point when the hype hits the reality and things really begin to change.

It looks likely AI could reach that point in 2025. If 2024 was the year of excited talk about what AI could do, 2025 could be the year in which the reality of what AI can do really starts to have an impact on our lives, moving from the realms of science fiction to science fact.

Once a critical mass of users get used to AI assistants doing their routine work tasks or – widely predicted as a retail breakthrough – doing the weekly shop, we will have reached the point we did a few years ago with smart phones.

And once we get to where we can’t remember quite how we lived without AI, things could get interesting.

AI dominates any discussion of the outlook for the tech sector.

Themes will be increasing safety and reliability, applying AI more widely to compliance and reporting tasks, grasping the detail of how to maximise the effectiveness of AI alongside human workers and developing robust regulatory strictures for the use of AI.

Companies will need to invest significantly to bridge the skills gap – there is a shortage of engineers with AI and data analysis skills.

And investors will be looking for clearer evidence of revenue generation potential.

Even with a softer regulatory approach to AI tools, the spread of deepfake pornography, political deepfakes and disinformation campaigns have caused widespread concern, and there either has to be true industry self-regulation here or government regulation, or both.

A “no laws and no regulations” (or even a “too-little”) approach – when these technologies are so broadly used already (and there are so many stakeholders involved) is far too risky.

And this is especially true since rampant unethical use will just hinder innovation – because technology depends on the trust we all have of it to work as intended, and preferably without it harming us.

AI use will continue to be adopted by financial services firms at an accelerated pace, especially in relation to analyzing internal data and combatting financial crime.

Proposed UK AI legislation will be far less comprehensive than the EU’s AI Act and we believe that it will focus on safety of frontier systems.

I am still an AI sceptic.

Not in the sense that I don’t see any promise in the technology, but because I believe that the actual capabilities of AI are being significantly overrated. The reason for this is very simple. It is in the interest of those building AI companies to heighten the euphoria.

In my view the next two or three years in this space will be all about over-promise and under-deliver.

Having said this I do believe that the technology is interesting in terms of creating a symbiotic relationship between a system more capable of introspective and rational analysis and the human operator utilizing it as a tool for specific tasks and objectives.

What I do find interesting is the re-emergence of logic as the key to building the technology of the future. Will another round in the bout between rationalism and empiricism follow?

Another thing that has also occurred to me recently is that there is real potential for this technology to be banned outright should it, for example, end up being used autonomously in a battlefield context. Is it not possible to envision such bans in countries of all sorts of ideological leanings if robots actually started killing people without being ordered to do so?

Will anyone be able to do anything without AI or particularly ChatGPT in the future? I personally don’t use it, but finds it scary how broad it’s been used, from anything from making up your team pub quiz name or to cheat in it and ghost write serious academic or professional reports.

I also found it frightening how broadly AI has been used around faking images and videos on social media – where a lot of people thinking what they see is it real – and we will sadly see more of that.

Sticking of the scary bits, AI will also continue to be used by hackers and cybercriminals to create more effective campaigns to defraud and to improve hacker tools.

Regulators continue to produce guidance and guardrails but the future remains somewhat undefined, and while this will change the world we live in (including in compliance), we are still some way from fully understanding the use cases which will affect all markets and the risk management consequences of it.

My expectation is that by the end of 2025 some embedded AI will be the norm for financial services compliance teams, though this will not be the end of the race by some margin!

2025 could be the year we truly wake up to the reality of AI, with advanced technologies such as GenAI set to play an even bigger role in how the world works.

But also, and more importantly, this could be the year we get to know more about artificial general intelligence, or AGIs, the sort of tech that could, potentially, outthink and overtake human intelligence.

The use of Gen AI in everday work is going to become mainstream, and while this will increase productivity, this will be to the detriment of the quality of the end product.

The use of AI as an interface in finance will be supervised closely to ensure that the outcomes for customers are in compliance with regulatory standards.

Regulators might consider making disclaimers obligatory so that a consumer is aware if their interaction is with a human or a bot/AI agent.

The regulation of AI at national levels will become an arms race where certain states will sacrifice safety/security in order to encourage innovation and develop a global lead in AI advancement.

Big Tech will continue to invest handsomely in AI and investors/analysts will look ever closer at when this will reap a decisive return on investment.

Little tech will struggle to fuel further growth as interest rates remain a burden to further capital from the private equity sector, as well as falling victim to the growing corporate demand for better security and governance in the finance sector.