Tag: chatgpt

The Future of AI Adoption: Trends and Predictions for 2025-2030

3 Comments

I have been speaking about the challenges and potential of AI adoption for over a year now. The latest deck from my session at the South Coast User Group in April is available here. During those 12 months, a lot has changed in the world of AI…

Recently I spoke to Markus Erlandsson and Malin Martens on the CRM Rocks podcast about all things AI, adoption and the future trends for AI in general (you can find the episode here). I promised I would create a post with all the useful resources I have found over the last few months – you can find these at the end of the post.

But first, the future of AI…(based on research by the wonderful people of Gartner).

2025

Here we are, in 2025 with at least 30% of GenAI projects being abandoned after POC due to one or more of the following:

  • poor data quality
  • inadequate risk controls
  • escalating costs
  • unclear business value

We are experimenting. With that experiment, a few things have to go in the bin because they aren’t working. And that’s ok! By the end of this year, 30% of enterprises will have implemented an AI augmented development strategy. They will also have implemented a testing strategy. Software has changed the world, AI is changing software.

This is also the year that AI Legislation has started coming into effect. The EU AI Act, which I wrote about previously, is primarily responsible. Although no one seems interested in this yet, in a year, this will have massively changed. More of the staggered enforcement will come into play. Read up on it now, and make sure you get your AI inventory in place…

2026

Next year will be even more data focused when it comes to AI. AI will drive personalized adaptive user interfaces in 30% of new apps, up from 5% today. Over 100 million humans will engage robocolleagues, or synthetic virtual colleagues, at work (apparently, although I personally am skeptical of this one).

Gartner also predicts that 75% of businesses will use GenAI to create synthetic customer data. This is very exciting! By 2030, it is expected that the majority of AI models will be using synthetic rather than real data. I am a big fan of synthetic data so if you read about one thing, make it this!

2027

The lawyers will be very busy in 2027. Gartner predicts that by then, 70% of new employee contracts will include clauses for licensing and fair use of their AI persona (AretiBot, anyone?). This will have huge implications on how we work. It will also affect ‘who’ we work with. Can your AI persona work at more than one organisation at once? What happens if the organisation is making money from your AI persona? What happens when you leave the organisation?

In other news, it is expected that more than 50% of GenAI models by then will be industry or business function specific (fewer than 1% are today). This shift will make them more precise and thus more useful.

It is also the year when nearly 15% of new applications will be generated by AI without a human in the loop. Is this a good thing? Responsible AI practices suggest otherwise. Remains to be seen…

2028

Things will get weird in 2028… All the stuff we are working on right now will by then be technical debt. Yep, AI tech moves fast! Gartner predicts that 50% of enterprises will stop using large AI models built from scratch. They will do this due to cost, complexity, and technical debt in their deployments. So if you are not thinking long term (which in the world of AI is 3 years!), well, do.

What is even more exciting (and equal amounts unsettling) is that machine customers will make 20% of human-readable digital storefronts (see websites) obsolete. By 2030, 20% of revenue will be from machine customers. Imagine your printer runs out of ink. It orders itself more ink. Your smart fridge recognises you are low on milk. It orders you more milk. A human won’t be doing the buying – an agent or ‘machine’ will be. This is fascinating. I am endlessly curious to see how marketers will try to ‘influence’ machines and algorithms in their purchasing decisions. Humans have complex decision making patterns, machines less so… I have no doubt they will try to encourage your printer to buy the official ink though… 😀

Please also note that by 2028, 75% of enterprise software engineers will truthfully say they are using AI coding assistants. In early 2023, less than 10% admitted to this. I think it is 80% right now but the stats will take a while to catch up!

2030

Five years from now the world of AI will look VERY different. Synthetic data will be all the rage for AI models. More importantly, us humans will engage with software in a different way.

Historically, humans have been central to the use of software. We tell it what to do. We move a mouse, type on a keyboard, and use our voice. But in the future, we are there on a need to know basis, by exception only. Agentic AI and autonomous agents will be everywhere. Some predict there will even be organisations with more agents than employees. I absolutely believe this. Moreover, guardian agents will oversee AI agent actions. This will be very much needed as people will inevitably try to get agents to go rogue. Let’s be honest, sometimes they go rogue on their own!

I will sit down and contemplate how I want my fridge to decide what food to buy for me. I will also think about how to override that decision when I really do want some chocolate for dessert.

Enjoy the links!

Useful Links

AI Governance/Strategy

AI Tools

Copilot

Accessibility

Responsible AI

General