The Impact Of AI In 2025
Every year I create a master list of important themes going into a new year, this year the list is longer and more in-depth than before.
Why? This is down to the shifts happening financially with companies, the external factors such as political change and unrest alongside the seismic shift that is AI. Many doubling down into AI and placing most of their chips into building on top of other people’s technologies.
This is a dedicated list of how I view and see AI impacting companies and, importantly, their teams.
This deepdive also includes some thought-provoking questions to help you consider the broader implications for society and cultures.
A few things to keep in mind with your customers
Crossing The Chasm
Almost all customers are Average Joes/Janes aka pragmatists - they don’t rush for new features and updates - they often wait and dislike change
Most buyers are looking for problems to solve months in advance, sell the future and what it’s likely going to look like and answer
Customers demand the best possible solution for their problem(s). Offering a bad experience to get their problems answered will hinder your company. The interface and chat experience (right answer vs multiple attempts at an answer or a solution) will be imperative
A large percentage of customers have hit their limit on accepting costs being placed on them and having to back pricing increases. With product switching and going from big brand to own brand - many customers won’t adopt big shifts or price hikes
Crypto-tech and fintech will have the potential for another breakout and record-breaking year with tech shifts, appreciation for their technology stacks and embracing tech shifts early
Experiences Please
New customer experiences will take a long time to adopt so plan carefully and build brilliant experiences that are for your core customers not rush for an interface where customers will struggle (chatbots are still a horrendous experience for most customers - despite what is being sold)
This might be the spiciest one - we will see more AI automation, and will there be more parasocial relationships driven by AI interactions? Likely - I can see many Onlyfans models turning to AI to power chats and personal connection creation. Many people could turn to have what they want when they want from creators and AI creations all at a cost, rightly or wrongly this might add to some people’s addictions.
With huge pushes into AI there will be the push back, curated experiences by humans, human-powered collaborations and human-centred movements. For all that embrace, there will be those who embrace human art(s).
Agents Vs Web & App
Many companies will struggle with creating a digital presence that will create company agents that could become the canonical experience. We could look at a long-term approach to supplement and replace large clunky websites and apps. Many customers will always prefer in-person shopping experiences and creating experiences with others not just digitally - customers will love or hate this
Could we land in a place where MR (mixed reality) is embraced more, AR could become more commonplace as we look for different modalities to entertain and engage us
AI will have to shift from agents to collaboration partners to be more trusted and be accepted in many businesses
AI Wrappers (tools) will be smoked out early and wrappers will come and go in record time
AI will have to push for real-time vs powered by training and scraped data - the implications might be harsh for many businesses
AI builders will have to focus on being for personal usage or team usage - the smarter builders might double down on personal tools
AI hardware will be released and not force fit to older devices - this will mean much more powerful models on devices (old device - old models or no models)
Entertainment
AI will create entertainment and will customers care if it is automated or will the shift be supported (think from social media platforms to entertainment providers like Disney, Netflix and Prime Video)
Health
Healthcare has the ability to be cheaper, to be more personalised and more adaptable at scale with AI - will healthcare companies and institutes allow this…
Where there is huge spending in food (weight loss), defence (armed forces/warfare etc) and geopolitics AI will be at the heart of much of this
Society
Big shifts in Media, big media looking for partnerships and revenues, independent journalists will have more to work with and break more stories - what influence will this fracturing have on customers and industries?
A big impact on broader society and building different norms, AI partners specifically girlfriends have become something that has bubbled under the surface and it will likely increase in popularity. Dating apps have failed many so why some will turn to an AI partner looking for connection?
TRUST
Verify Then Trust Vs Trust Then Verify - a learning curve that will have to be adopted by everyone, we will have to verify the information, media and news before trusting, whereas currently most trust then verify.
This is a phrase that Scott Belsky coined and has rightly been pushing and integrating into software and apps, many others aren’t.All In & All Out - there are always those who go all in and those who reject new tech. AI to many seems complicated, a new language and to some evil. Some customers will go all out, especially with the pressure on energy to power data centres and the environmental costs.
Company Agents - digital presence is essential for brands. We have seen website functionality merge and become easy-to-recycle product features, the best apps have improved experiences that surpass web, more than just for convenience and a number of important notifications. It has led to huge spending and a huge portion of people’s waking hours on their smartphones. Companies will consider having to dive into creating company agents that interface, converse and recommend solutions or products directly to their customers. AI company agents could replace long boring onboarding flows - helping customers to install apps and software and handle customer support and success experiences all within apps like WhatsApp (OpenAI’s chat in WhatsApp is a smart way to distribute and add more convenience), iMessage or a new centralised chatbot space. How will you stand out in another tech space?
Trust is always critically important whether you are an AI company, a company leveraging AI features from other LLMs and companies who will want to be ahead of their competitors and will want to cut corners. Always bake trust into your experiences and enable customers to opt in or out of features and control their data. AI will be scary for many and will lead to numerous second and third-order effects most Product and safety teams won’t have time to consider.
UI
Distribution Of AI Content - The way we consume content has shifted, the best content will be distributed to the customers on the platforms and apps they select, AI content will need the best distribution to cut through. Easier content generation means a flooded market and more Marketing - it’s not always better or more effective Marketing of content. Understand how you deliver higher quality not higher quantity. The interface of how people consume your content will be critical and it will likely be as unbundled and uncentralised or worse over the next 18 months
Learning Prompts Not Just Searching To Find Solutions - right now we are operating in a chat (ChatGPT or Claude) or search (Google or Perplexity) interface, with many needing long complicated prompts to answer customers' questions and visions. The best products will reduce the friction of learning more long prompts and evolving prompt libraries to become more user-friendly (artefacts is a start) and help ask questions (improve your prompts without you knowing) to gain better answers. Many customers want peace of mind not always a solution.
Remember a Product has to be (1) easy to learn (2) easy to use (3) easy to share (4) easy to personalise.Chatbots As The Primary Interface - learning how to ensure Chatbots work as well as possible or will chat interfaces have to change quickly to get the most adoption. There will be a moment when chat interfaces have to change or evolve from chat box to something else to create a rounded experience (so many users don’t know how to get the best results out of Google let alone a chat interface)
“Agents” - We have seen assistants be renamed to agents and how agents will go off and do activities for us from review and package the news → to find the best price for flights (basically replacing mass searches and using price comparisons etc). Agents will become an interface we become familiar with but the test of time will be the results and not using the same few websites or companies as their default answer set. Agents have huge potential but curating news, listening to podcasts and watching key YouTube videos will only be as good as the information we take in and learn from.
Integration Into Everyday Apps - the smartest way to distribute AI will be to live within the most important and popular apps. Whether that's inside chat or inboxes or be a constant interface. Increased usage will be critical to improving AI and specific personalised experiences. Without deep and constant integration most experiences won’t improve and likely won’t improve the success rate of answers and reducing hallucinations.
…Or Into Operating Systems - Microsoft, Apple and (Google) Android have a risk-reward play here, do they bake into their OS’ and push for usage to sit above apps and websites or do they partner up and allow many to offer questionable experiences. Apple might well be best placed to push ahead (more than just writing tools and image creation) as they do not have hundreds of partners and vendors all competing with their own devices like Google and Android and Apple will land in a place to connect their ecosystem - despite the anti-trust investigations.
The UI of AI is the battlefield that someone has to win to rethink the solutions they are offering. More users means more interaction data however usage data won’t help if you are offering a solution that doesn’t work or won’t work for most users. Modern-day AI is effectively in year 3 so adoption will increase but so will users deciding to be in or out of dedicated shifts.
CONTENT
Generation Of Images And Videos And Autogeneration - we will see huge amounts of content created and shared online. We will see a large variety of quality and whilst it might seem trivial, the quality bar of acceptable will be incredibly important. Brands shouldn’t look to generate and autogenerate large volumes - algorithmically most accounts across social and paid platforms will reduce the reach quickly of low-performing content
Personalised Content Created By Platforms - Something we might see is how Meta and other platforms start to create hyper-personalised AI content based on your behaviour if engagement slows from your curated network and from newsfeeds. Engagement is key to these platform successes and with such low friction it would be easy to create personalised every time you log in / open the app
Personalised Emojis - consider how hundreds or thousands of genmoji creations will cause questions and confusion from user to user. Thousands of new personalised emojis (aka stickers) being used will be entertaining but will also mean many ask what emojis mean away or cause offence across different platforms
Low-Quality Junk - the quality of most AI-generated images is low and people are ok with that. The quality of web-written content is still relatively low quality and formulaic - some companies have already decided the ease of creating low quality is worth risking for their SEO. The demand from most companies will be to create more (more will seem too easy not to push on), it will cause a tsunami of low-quality content being uploaded and shared and most customers won’t entertain it.
More Aggressive Filtering - More content and easier-to-create content (even automated based on user queries) will lead to more aggressive filtering of content that lands in your feeds, inboxes and mobile notifications. More filtering is inevitable and will greatly impact the sizes of engaged audiences (think social media, email inbox filtering and notification reduction from Apple)
High-Quality Work Is Seen As AI Or As Assisted By AI - One major issue we will encounter is will all high-quality work in time be seen as AI or be seen as assisted by AI, the risk for many is giving credit to AI when it wasn’t AI it was assisted by AI. Many companies will look to hire and fire based on output and performance, more that is seen as AI only the more it will give senior leadership teams disconnected to perform “operational efficiencies”. People hate colleagues who steal ideas imagine how team members will react when AI is given the props to great campaign or product features instead...
Generative Art On The Blockchain. Going Beyond Bad NFT Experiences. Super high-end art being created and then becoming personalised to you specifically, not just one version of one or 10 of 10 prints etc. Allowing smart contracts and IDs to be used, verified and leveraged. The possibility is huge if managed correctly. Maybe
AI Voice Improvements - turning more written into audio (read company memos in CEO voice, read company emails and comms into voice - make it authentic and in the medium people want to consume) and being able to add more personality and resonate with the leaders comms. An issue with improved AI voice will be the ability to cut through the noise of more comms not better comms (more mediums = more noise and notifications). We can also see how Podcasts and radio could be improved and deeper connections can be built between podcast host, radio host and their listeners. AI could be the force behind turning more fans into superfans…
AI Influencers On The Rise - we have seen some resistance to the Puma influencer but there are many doing incredibly well and can look and feel very different with a prompt. It can go from villain to hero (or vice versa) very quickly and scale with direct relationships with their audiences. Will companies invest teams into building communities behind AI influencers? It’s possible, but not impossible and can test outfits (skins) without having to create physical copies. The first to crack this new approach will have an interesting challenge to work through signal vs noise.
Long Form To Short Form - Short Form To Mid Form Content - what is exciting to most people will be the ability to leverage AI to create numerous great content packs from long form videos and longer form content types and then how you can turn a clip into mid form (say a minute to 10 minutes). Think of great clips and Notebook LM blended together on AI steriods.
Decide at what end of the scale you are Good Enough → Good → Better → Best → Greatest, if you are deciding Good Enough, expect issues with platforms reviewing your rankings and how you appear in their feeds or inboxes…
Many executives will struggle to understand if AI-generated mascots, characters and influencers are right or best for their business.
IMPROVING WORK
Workers Adapting From Do-Ing To Reviewing/Editor Role AI Work - the upcoming shift will mean many will have to adapt from doing the work to reviewing the work which will take large shifts and retraining will be needed. There will be a case where automation of work do-ing to reviewing and editing will either reduce headcount or increase demands on re-viewing and editing which many aren’t skilled to undertake
Internal Tool Builds Will Be Essential - building your own internal tools and personal LLMs will be critically important to get ahead and improve ways of working and compete against competitors and new entrances into your market. The best companies will be building internal tools in the very near future and it will be unique to those internal vs being a tool that becomes widely available and plug and play for your competitors
Internal Chatbots Powering Internal Knowledge Sharing - the flow of knowledge sharing and keeping everyone (and their “assistants”) up to date and empowered will be critical and where some will win over many of their competitors and market leaders. Notion got their implementation nearly right, will companies be confident in being as precise and thoughtful in improving their internal comms and knowledge sharing? I hope so
AI HR Issues - I have a concern that a handful of people will try and shortcut a number of management and colleague-related issues with AI and it will cause serious offence, as you will not be able to input enough context and touch points for AI to write something for you or create something that won’t create issues for people who copy and paste answers. There are a number of AI driven HR tools that will make important decisions based on data that isn’t up to date, isn’t complete and does not understand nuance and relationships enough. Will HR teams and leadership teams be ready to explain hiring and firing decisions based on a HR 9 box exercise or a scoring matrix they do not know all the parameters?
AI has the ability to make or break, many companies are focusing on external tools and building more rather than improving their internal tooling and processing. Cross-functionally AI tools have huge potential - it will be how exec teams manage this and encourage the right build of tools rather than renting a tool that all their competitors are using.
CONSOLIDATION
Consolidation Of Tools And Apps - this is already happening so won’t be a surprise but so many tools and apps are built on the same tech, the quicker they update and slower adaption happens many will actively select to consolidate to continue to operate
Device Side AI - We will see many devices take the load on the device away from the cloud and this will be good for the user until models are out of date or not updated for weeks vs in the cloud.
Sherlocking Of AI features & Apps Into Claude & ChatGPT - We have seen this in the mobile app space we are likely to see this on Apple and Android devices where AI features will be baked into the OS and apps won’t be needed. We have seen this in OpenAI already, it will be a matter of time before Apple, Google and Microsoft attempt similar in their OS’ and device side with an app.
Consolidation is inevitable, we have seen consolidation already happen. Through 25 & 26 we will see way more consolidation and competitors join forces to save their companies. We will see many agencies consolidate and merge and some fierce competitors will look to join forces to win the headwinds coming from the AI battles.
OPERATIONS & PRICING
The Demand For ROI And Overspending On Nvidia Chips - there will be challenging times when companies and PE/VC are pushed on returns and will see many companies running leaner or shutting their doors much quicker
Gemini’s Pricing - Google AI has to compete in smarter ways not just in search and advertising but across their office suite. Google to stay relevant could hugely discount Gemini features and open it up to every Android user or every Google Workspace user for almost cost prices. Gemini pricing could be key to competing against early AI leaders rather than thinking about their ads driven business.
The AI PR Problem - it is going to happen. The AI trust scale will come to light and consumer trust and PR problems will bubble up. Where AI generates a series of questionable content, and automates a build for bad apps or tools that exploit people there will be a huge PR problem for these companies and the tech that these are built on will be guilty by association. Something that will be increasingly important is how these companies handle bad press and ongoing bugs and how they interact with the public. We have seen many examples of woke AI, AI filters that have influenced beauty trends and some influencing self-harm, many are going to have to work much harder to prevent users struggling to understand the implications and how they handle responses. (character.ai example from 2024)
AI is a pay-to-try movement, credits are often wasted and can feel like burning money. Most of AI tools aren’t fit for purpose and are often small tweaks of other tools. 2025 is another year with more with less and budgets are going to be tight, chasing more tools won’t be tolerated and ROI will be pressed. In Q1 into H2 we will see a lot of churn and closing of AI tools and the big incumbents will decide what their finance risk tolerances are and if they invest into Nuclear and if they continue investing billions into their platforms. Meta, Alphabet and Microsoft will be worth watching throughout the year.
Be very careful in the HR & PR problems AI will bring.
SOCIETY & CULTURE
The Speed-Up Or Slow-Down Of AI—AI experts (and researchers, etc.) are pushing two very different agendas. Some are pushing the speed-up and acceleration of learning, while others are pushing the narrative that it will slow down and is not progressing as quickly as many hoped. The over-investment by some will bite hard.
AI Becoming Normal - accelerated chasing the chasm by integration into mobile operating systems. The use of AI will be more seamless and often unknown to users. Marketing with AI will continue for many brands but until products improve in solving problems or adding more time or money consumers won’t likely select tools just because of an AI tag or tagline
Costs Of AI On Young Members Of Teams & Kids - one area that is being under-discussed is the cost of AI on young members of teams and on children. Children are adaptable and will use like another tool, younger members of teams will look for cheats, shortcuts and for AI to complete large sections of their jobs. Unless they are taught the issues with relying on AI and understand the second and third effects many may struggle after quick hacks or shortcuts. Managers and company leaders have to be aware of this shift and be able to guard their company and their employees from over-reliance/non-disclosing of the use of AI.
AI Leaks - Leaks are “normal”, what might become a mainstay is the future of features or products and leaked with AI or by AI. AI features are being oversold currently but leaking features and upcoming big launches are big business, what will be important is how you understand leaks and any implications. AI-driven leaks might be early but it is something worth considering when teams will rely on AI to create ads, creating content, creating apps etc.
Age Verification - some tools will have to consider more aggressive age gating and limiting users by adding tighter age restrictions. Many AI apps will be under pressure for users and monetising users whilst others will be more concerned with long-term revenue and success and protecting users will be more important. Age verification will be a critical part of protecting users.
SECURITY
AI-based Hacks - Some good some bad - white, grey and black hat hacks will be viewed as bad by default. Some will question the influence of AI hacking, others won’t. The bigger issue that will be overlooked is how long it will take to feel any effect from these hacks. AWS have reported numerous AI hack attempts and this will only increase with tools like Replit creating and re-producing websites, apps and tools in minutes
How Companies Will Attempt To Manage Good AI And Police Bad AI Practices - this is an area we have yet to fully see roll out and many companies are struggling to understand and then police AI usage and the data that is shared. Something to keep an eye on is the shear number of requests to log in with Google and the access levels provided to real-time and historical data
Software Brought In By Employees - we saw with Dropbox, Slack and Figma how software and apps came from staff not IT, we are seeing so many tools being tested on company data it will be a nightmare to manage from a security perspective. AI has huge potential for good and bad - often team members are looking to jump to an answer or get assistance on an impossible goal or task - sometimes decisions
Copyright Challenges - the impact of the court cases around copyright and ignoring robots.txt file(s) and scraping data will be essential to keep an eye on and understand the implications on your business. There will be many issues with image and video creation and filters used let alone intelligence gathered on your data.
The AI culture wars - will it be good for society or incredibly bad for society? This is going to be an ongoing question for years to come, with tribalism already dominating Western culture we are going to see more content and more influence from bad actors. Will countries come together to control their influence on culture or will it fracture culture even more than today?
The old saying of you won’t be fired for hiring IBM, maybe the next phase is - you won’t be fired for letting AI decide (for you)…I hope not but many are getting to this point.
2025 has huge potential with AI - it’s on us to educate, train, think more critically and police in the right ways!
Dive Deeper On AI
By Me
By Experts
