--- title: "Where AI is really at: Q4 2025 Reality Check" url: https://sam-page.com/thinking/where-ai-is-really-at-q4-2025-reality-check date: 2025-10-15 author: Sam Page type: Post or article categories: [Emerging tech and AI] reading_time: 5 min summary: "Over the past 3 months I’ve spent lots time speaking to people across many companies, including global brands (FTSE+NASDAQ100), scale ups, finance and tech firms and purpose-led smaller…" first_published: https://www.linkedin.com/pulse/where-ai-really-q4-2025-reality-check-sam-page-ejfme --- # Where AI is really at: Q4 2025 Reality Check Over the past 3 months I’ve spent lots time speaking to people across many companies, including global brands (FTSE+NASDAQ100), scale ups, finance and tech firms and purpose-led smaller businesses, at all levels of roles, from C-suite to implementation teams. I've been offering advice, but at the same time, learning what really is and isn’t working for them when it comes to using AI – in particular its role in marketing, customer and digital experience. After all, deploying creative technology needs to be a collaborative process not just a presentation or framework!!! One thing that has struck me throughout all the conversations, is how similar they have been, regardless of industry and company size. **Some important context** I've been hands on with AI at [7DOTS](https://www.7dots.com) for the past couple of years. We've been building our own tools that are AI enhanced, helping clients to deploy AI where it can create an advantage, and launching an AI joint venture called [DealCraft](https://www.dealcraft.ai/) – a sales enablement platform. We have a lot of practical experience. **What am I finding out about AI adoption across companies?** Firstly, individuals at all levels are using AI more, both at home and at work. Not only to ask questions, but to help them make decisions, including what to buy. No real surprise and this is backed up by numerous studies! At a company level, brands are trying to adopt AI based on the promises of it revolutionising the workplace, fear of missing out, fear of messing up, and often as a cost-cutting exercise. However, the results are often very disappointing – the main thing that people are telling me is that quality dips and creativity is lost. > “Everything becomes a bit average” As a handful of people have said to me. And to compound matters, the cost of deploying and using AI is surprisingly expensive. It is akin to the “[enshittification](https://en.wikipedia.org/wiki/Enshittification)” principle and in part a result of it running out of original content. Something [Ed Gossage](https://www.linkedin.com/in/edgossage/) wrote about in 2023 in his “[Ai will eat itself](https://techround.co.uk/interviews/ai-will-eat-itself/)” article. Yes, there may be some bottom-line savings, but the negative impact on the top-line is counteracting them. In short, lower standout, less engagement, lower sales figures. Replacing people entirely, or using AI to cut corners, isn’t working well. And although many boards are actively focussed on cost-cutting, there is a real sense they will start to see the negative impact on the top line soon. [Klarna’s recent roll back](https://www.forbes.com/sites/quickerbettertech/2025/05/18/business-tech-news-klarna-reverses-on-ai-says-customers-like-talking-to-people/) on replacing customer service advisors entirely with AI is a good example of this. A recent article in [Marketing Week](https://www.marketingweek.com/ai-powered-personalisation/) summarised it well – identifying that only 12% of AI projects are delivering a measurable ROI. **Beware of AI snake oil salesman** Of course it’s not helped by the number of “AI gurus”, claiming to revolutionise businesses, and save them millions, overnight, from a series of white-papers and workshops alone (which they are charging vast sums for). Ironically most of this is AI generated, which is why it all looks the same on LinkedIn... and perhaps why so many people are fed up of LinkedIn at the same time. 🫠 In other words.... If it sounds too good to be true, it is, otherwise everyone would be doing it. **The over-reliance on AI by team members** The other challenge I have heard time and time again, is the emerging over-reliance on AI by team members. Which is leading to poor quality output and mistakes – as demonstrated in [Deloitte’s recent report for the Australian Government](https://www.theguardian.com/australia-news/2025/oct/06/deloitte-to-pay-money-back-to-albanese-government-after-using-ai-in-440000-report). They are refunding the government in part due to the errors in the $440k report. It’s critical to remember Gen AI in it's current form is only 90% accurate (at best), and isn’t capable of true strategic thinking, or actual artificial intelligence. Well demonstrated when [Magnus Carlsen beat ChatGPT at chess without losing a single piece](https://time.com/7303017/magnus-carlsen-chatgpt-ai-chess/). AI should inform, with a healthy level of human challenge, rather than dictate. **In summary, hype is turning into reality** As we move towards the end of 2025, it’s becoming clearer that AI is not the answer to all our problems, as the hype would suggest. At least not in its current form. **The gap between expectation and outcome remains wide, driven by misplaced priorities, incorrect approach, poor implementation, and an overestimation of what today’s systems can truly achieve.** **But it’s not all doom and gloom!** I hasten to add, I am a huge proponent of new technology, including machine learning (ML) (which has been around for a long time) and Generative AI. And combining the two. When approached in the right way, and used correctly, Ai can give **both companies and people a huge advantage**. **I'm seeing and experiencing this first hand as well.** However, in an attempt to keep this article relatively short, I will explore these in **Part 2**, next week, **along with what I’m seeing on the horizon for large language models and the next wave of AI evolution.** **It certainly seems that the hype about models getting better (achieving super intelligence) is fading, and the big tech firms are starting to look sideways instead....** more to follow. Original image by Google DeepMind: