What to expect from AI in 2025

An image of a network in the shape of a brain representing artificial intelligence.

What to expect from AI in 2025

An image of a network in the shape of a brain representing artificial intelligence.

Artificial Intelligence is seemingly the #1 topic of conversation in every whitepaper, conference and boardroom right now. At Reputation Leaders, we’ve seen a convergence of opinions that offer insights about what to expect from AI in 2025 and how to plan.

1. AI is not equal opportunity! It’s likely to widen inequalities between the haves and the have-nots.

The IMF has said that AI will affect almost 40 % of jobs worldwide, replacing some and complementing others. However, they estimate the number of impacted jobs in advanced economies is more like 60%, while in low-income economies, only 26% of jobs will be impacted. (Let’s take a breath and think about the fact that A QUARTER OF ALL JOBS seems like a small amount in comparison!)

Advanced economies with the infrastructure, technology and educated workforce are likely to pioneer AI and ultimately guide its usage.

Within countries, skilled workers who can harness AI to increase productivity are likely to see their wages increase along with their opportunities.

Large companies with capital to spend on higher wages for skilled workers, AI infrastructure, and the ability to find economies of scale are likely to reap the most rewards from AI.

The AI gap could exacerbate inequality of other kinds and lead to social unrest. The world needs protection against this trend.

Recommendation: Assess how AI is likely to impact your job, organisation and country given the infrastructure and skills available. Plan to put in place as much future facing infrastructure as possible. Increase digital access and access to AI globally so that everyone can benefit from the AI revolution.

2. Opportunities for disruption. Large players may be more resistant to change, while agile AI adopters can seize advantage in the face of risk.

AI can bring the power and productivity of a large group of specialists to a smaller group of generalists. AI can now create images, write copy, give information, and interact with the world in a way that is often good enough, particularly when it’s used in collaboration with human thinking and efforts. The same applies to image creation, policy drafting, and data analysis, to name a few.

This means that agile players, willing to pivot and risk the output being good enough, are in a prime position to disrupt the market, whichever market they are in. That’s true for individuals within companies, companies within industries, and countries within the global economy.

Opportunities come with risk. Established players are less likely to take risks and can miss opportunities that agile risk-takers will likely take advantage of. Consider the Kodak case study from the 1990s and early 2000s.

Agile adopters have the opportunity to disrupt established players and create a reshuffle in some industries. However, consumers should be cautious, as not all of those risks will pay off, and some of them could put customers at risk.

Recommendation: Openness to new tools for work is essential to stay competitive while recognising risks will be crucial. Be cautiously optimistic. Fast followers of successful AI use cases can adopt proven AI quickly, without putting all their eggs in one basket.

3. Hyper-personalisation of services, entertainment, work and anything else you can think of.

We’re really excited about a new product called Inca. When we get the results from a survey (we run a lot of surveys!) we always look for the places where respondents could comment or explain why they answered the way they did. We love understanding what motivates people! Unfortunately, it can be challenging to analyse because sometimes people don’t write much.

Inca uses AI to respond to a person’s comment in real-time and ask personalised follow-up questions to understand more. We hope this hyper-personalisation will encourage people to respond more fully to questions and help us understand more deeply the thinking behind their responses.

Hyper-personalisation is coming to the market research industry. It’s already driving entertainment with Netflix and Spotify creating your personal playlists.

McKinsey called out industries that would be significantly impacted in their 2024 Vivatech presentation, and it’s easy to see how AI-powered hyper-personalisation will impact financial planning, insurance, healthcare, shopping and others.

But with hyper-personalisation comes hyper-privacy concerns. Individuals will need protection to make sure their privacy, both physically and digitally, is protected.

Recommendation: As much as possible, help AI to react to direct inputs in personal ways, but don’t use information that people haven’t provided directly! That’s just creepy.

4. New information becomes more valuable as existing information is available to everyone.

We currently have more data than we can use. Hitachi Vantara’s report “Drowning in Data” shared that large companies expect to store an average of 65 Petabytes of data by the end of 2025. Don’t worry if you don’t know how much 65 Petabytes of data is. Nobody else can either. It’s unfathomable.

AI-powered systems are making that data available as summaries, search results, and visualisations in ways we’ve never seen before. One of the most exciting ways AI uses existing data is to create synthetic data, which uses what we already know but blends it to achieve new insights.

A common criticism of synthetic data, however, is that it can only tell you what we already know.

What makes synthetic data exciting, is the possibility to generate that data in ways that can be merged with new primary data to generate new insights. For this reason, primary, proprietary data that you can add to the existing knowledgebase to generate truly new insights unique to your thought leadership will become even more valuable.

However, by reusing and repurposing what we already know, the existing biases and thought patterns risk being reinforced and coded into this synthetic data. Historical victims of bias need protection against the proliferation of bias in our digital knowledge base through the scrutiny of our data and data collection processes.

Recommendation: Connect models with proprietary data to create more value, but protect your proprietary data more than ever.

5. AI Regulation will need to rapidly evolve to protect the public from the negative side effects of AI in everyday life.

The first four expectations for AI all have risks. There are people who need to be protected from the unintended impact of AI gone right or gone wrong. You are likely one of those people. So are we.

Both the IMF paper on AI and the Institute of Directors report prominently state that AI needs regulations and frameworks to protect the vulnerable and make AI work for everyone, not just the privileged few.

Expect to see a flurry of AI policies, guidelines, frameworks and regulations being produced by everyone from the smallest company for their employees to the international organisations representing billions of people. Navigating the regulatory landscape of AI may become more complex than the technology.

Recommendation: Stay up to date on the latest AI legislation news, as it could affect the trajectory of your AI use.