✨ Collective Predictions 2025
As 2024 draws to close, we asked each of the Time Under Tension team and Collective members for their predictions of what generative AI has in store for us in 2025. The predictions spanned 5 key themes across Business Awakening, Shopping Habits, Agents - Assemble!, AI’s Hidden Costs and Platform Saturation.
Full copies of the predictions below.
Theme 1: Business Awakening
HIT ACCELERATE!
I used to love this quote from the early Digital era … "The big don’t beat the small anymore, the fast beat the slow.”
Well that era is feeling almost sluggish compared to these Artificial Intelligence days, and in 2025 it is going to hit warp speed. Senior execs have seen the cliff-fall graphs of major corporations impacted by AI and they’ll realise that it is ‘change or die’. They will have to cannibalise legacy products, embrace the new, invest fast and build fundamentally different operating models to cope, thrive and innovate. Time flies, so be a pilot.
SMBs will increasingly adopt generative AI to personalise content, automate tasks, and, most importantly, reduce costs. A key challenge will be balancing autonomous decision-making (agentic AI) with the need for explainability to ensure transparency, accountability, and privacy. While fully explaining AI’s decisions is currently impossible, new tools leveraging multiple models will emerge to help manage risks and meet new regulations. At the same time, large AI companies will continue creating walled gardens, complicating adoption by pushing proprietary products at rising prices. Meanwhile, predictions of mass adoption of smart fridges with AI will be pushed to 2026.
I believe 2024 has been a year where everyone has started to take note of GenAI and the impact it can have on their businesses. I believe that for 2025, many sectors are poised to undergo a transformative shift driven by generative AI technologies. More businesses will start leveraging AI-powered personalisation creating hyper-tailored content that dynamically adjusts to individual preferences, past history, and real-time personal data. We will also see the growth of GenAI-driven virtual shopping assistants that provide immersive, context-aware product suggestions, enabling customers to visualise products in their personal spaces through advanced augmented reality interfaces.
These AI systems will not only enhance customer experiences by predicting precise needs and preferences but also optimize inventory management, supply chain logistics, and predictive pricing strategies. I also expect more organisations to integrate multilingual AI chatbots that can handle complex customer interactions with unprecedented natural language understanding, potentially reducing operational costs while simultaneously improving customer satisfaction across both online and physical channels. Gen AI will be everywhere as companies who have sat on the fence begin implementing GenAI capabilities leveraging the technology platforms who will provide these offerings out of the box.
Theme 2: Shopping Habits
ChatGPT launch an AI Shopping Assistant
Two weeks ago, we saw the first of the major AI chatbots launch AI Assisted Shopping. Perplexity Shopping now allows Pro subscribers in the US to ask for product recommendations and then buy them with one click from the chat (with card on file).
I think that in 2025 OpenAI will do the same.
If ChatGPT were able to nab just 1% of US eCommerce market share (Amazon has 37.6%) it could add up to $17 billion of revenue by 2028. That's a big prize!
With ChatGPT's brand awareness, market penetration and unrivalled access to data, why wouldn't they do this?
SEO in 2025: The Key to Unlocking Generative AI’s Potential
Gen AI is reshaping SEO practices by emphasising content quality, natural language alignment, structured data, and user intent. These adjustments ensure AI models deliver more relevant and user-friendly responses, reflecting a shift in how digital content is crafted and optimised. As AI increasingly powers search engines and user interactions, businesses will adapt by creating conversational, authoritative, and well-structured content that AI can easily interpret and prioritise. This evolution positions SEO not just as a tool for ranking higher in search results but as a foundational strategy for influencing AI-driven outputs.
Heading into 2025, the nexus of SEO and generative AI will significantly impact how businesses attract and engage audiences. AI models will favour content optimised for nuanced user intent and conversational queries, creating opportunities for organisations to build deeper connections with their audiences. Structured data, fresh updates, and AI-driven content creation tools will further accelerate this transformation, marking the convergence of traditional SEO and AI capabilities as a critical factor for staying competitive in an increasingly AI-powered digital landscape.
One other interesting and interrelated points to make is that many of the related strategies and tactics that will be used in influencing SEO to translate into better GenAI results, will be created by GenAI itself, meaning the potential for an endlessly recurring loop - insight / create / optimise / analyse / adjust / publish - could be a reality soon? How will GenAI choose what to put forward if 100% of these strategies and tactics were AI driven?
Theme 3: Agents, assemble!
In 2025, AI agents will transform how we work and live, with rapid adoption driven by their ability to streamline tasks, enhance decision-making, and provide personalized support. These agents will evolve from niche tools to indispensable virtual collaborators, assisting professionals with everything from project management to creative brainstorming. Businesses will embrace AI agents to boost productivity and reduce operational costs, while individuals will leverage them for personal organization and learning. In the creative services industry AI agents will also act as brainstorming partners, suggesting headlines and visuals, freeing up humans to work more conceptually. With advancements in natural language processing and user-friendly interfaces, AI agents will become intuitive, proactive, and increasingly human-like, fostering a new era of collaboration between people and technology and giving humans more freedom from the mundane.
What if AI knew EVERYTHING about you and could pre-empt your every move (in a helpful, not creepy way!). Drafting responses to messages, in your tone, exactly how you’d write them. Managing your diary, being aware of every personal and business meeting you have. Collating news, articles and products that you’d be interested in. A true personal assistant.
There is only one company in the world that can achieve this vision, with relative ease – Apple.
Every iPhone user in the world manages their life on their phone. All it would make “Her” (movie) a reality would be one permission toggle and an on-board AI (no cloud or internet needed) would get to work, becoming completely tailored to your needs.
So, my prediction is that Apple launches a next gen, LLM enabled, on-device assistant (and their share price doubles when that happens).
Looking ahead to 2025, AI agents are set to reshape how we interact with technology and conduct business. While there's a lot of buzz around these agents becoming our personal negotiators with brands and services, the reality will likely be more nuanced. We'll probably see them start with simple, standardized negotiations like hotel bookings or bulk purchases, where the parameters are clear and measurable. They'll excel at email management, calendar coordination, and home automation, learning our preferences and patterns over time.
The real game-changer will be how these agents work together in coordinated systems. Think of them as a team of specialized assistants, each handling different aspects of our digital lives. But it won't all be smooth sailing – we'll grapple with trust issues, legal frameworks, and security concerns, especially as these agents handle more sensitive tasks and financial transactions. Companies will need to adapt, building agent-ready infrastructure and APIs, while traditional customer service roles will evolve to work alongside these digital helpers. The transition will be gradual, with tech-savvy users leading the way before mainstream adoption takes hold. What fascinates me most is how this might democratize negotiation power – imagine having a skilled digital negotiator in your corner, working to get you better deals around the clock.
Here's a great example of how one guy has started creating voice clones to interact with the world in hilarious ways, obviously just the start of where this sort of technology will go, but a hint at what might be coming: https://davidepstein.substack.com/p/attack-of-the-ai-voice-clones
So what are the implications if you're a brand? Well, you may have thought you would be building AI Agents that your customers would engage with to save you time, money and effort in providing customer service or sales, but actually, it;'s likely to be your customers who are building personal agents that are tireless in trying to get a better deal, and will negotiate on their behalf until they find organisations who are willing to meet their demands.
Imagine an army of AI Voice Agents who are created by and working on your customers' behalf calling up your contact centre again and again, checking the terms of all of your contracts with providers, finding better deals, checking the fine print for service penalties, or asking for onerous clauses to be removed. Is your organisation ready for a barrage of customer agents asking for a better deal?
Theme 4: AI's Hidden Costs
2025 will be the year of AI ubiquity. AI tech will permeate every area of media and marketing and we’ll see great AI-powered creativity and productivity gains; consequently there will be ongoing disruption to job roles and organisational structures - those wielding AI tools effectively will see significant competitive advantages.
Australians are among the most leaned-in users of AI tools in the world, and we’ll see big strides in semi-autonomous AI agents deployed by brands to handle customer service, transactions and more, as genAI face and voices improve.
AI slop will contaminate the internet at greater and greater rates. Less a prediction than a hope, but 2025 needs to be the year of appropriate government regulation around AI, particularly in addressing deepfakes and disinformation - especially important in a federal election year.
In 2025 we will see landmark court rulings which will change the way AI trains on what is considered public data, the core issue being intellectual property. This will likely be specific to certain countries and will affect the knowledge that an LLM can access depending on a users’ location.
As of December 2024, over 200 million people use ChatGPT weekly - double the number from last year. Generative AI has become a creative lifeline, empowering millions to produce ideas, designs, and solutions. Yet, this revolution comes with a hidden cost: energy.
Each AI-generated image uses as much energy as charging a smartphone. Multiply that by billions of creative outputs daily, and the environmental toll becomes undeniable. Generating a thousand images is equivalent to driving a petrol-powered car four miles. Always-on AI agents, designed to provide constant support, add to this burden. OpenAI’s launch of Sora - a groundbreaking text-to-video AI - unleashes unprecedented creative potential, but its energy demands will undoubtedly push the limits even further.
The challenge isn’t just technical—it’s behavioural. AI doesn’t waste energy; we do. Every vague prompt, unnecessary rework, or aimless query burns power. The solution starts with precision. By asking smarter questions and refining prompts, we can cut waste, conserve energy, and achieve better results.
Yet smarter use alone isn’t enough. Companies like Microsoft, Meta, and Google are powering their data centres with renewable and nuclear energy. Wind, solar, and hydrogen are becoming part of the mix, alongside improved transparency. Leaders like OpenAI are already disclosing energy metrics to set new accountability standards.
Generative AI is a creative revolution, but it needs a sustainable framework to thrive. Companies should consider a creative offset: for every 1,000 prompts or images generated, they could plant trees, fund renewable energy projects, or invest in carbon capture technologies. Paired with energy-efficient models, smarter usage practices, and greater transparency, we can redefine what it means to create responsibly.
This is a chance to align creativity with sustainability—a moment to lead by example, turning innovation into a force that inspires while safeguarding our planet. Every idea, every prompt, is a choice. Let’s make sure our choices move us forward—responsibly, boldly, and brilliantly.
The Good:
As better, novice-friendly AI video/image/audio/writing/design/ideation tools are rolled out, we’ll redefine what it means to be ‘ a creative’ at work. With the right guidance, a gardener might produce time-lapse plant care guides. A kindergarten teacher could give her kids personalised interactive books where they’re the walking, talking characters in their own artwork. Teams of AI agents won’t just take care of grunt work, they’ll provide users with decades of knowledge, processes and frameworks that they can blend with their own expertise and experience. Millions of people who never considered themselves creative, will get to invent and make things they never imagined possible.
The Bad:
Advanced AI model prices will rise significantly, giving the Haves an insurmountable advantage over the Have Nots. Success or failure will increasingly be determined by whether a business can afford premium-priced models.
Theme 5: Model Saturation
The more we as an industry innovate, the more we will realise we don't need to. Companies like OpenAI may release further models, but I think they'll come to realise that we've already moved mountains as an industry, and the focus now should be adoption - not innovation. Don't get me wrong, I'm not calling for them to stop innovating entirely. But if we truly want AI to be the fourth industrial revolution - the next internet-level transformation - then they should be focusing on driving adoption, scaling real-world use cases, and embedding AI into enterprise workflows.
If OpenAI and other Gen AI giants continue releasing models without focusing on allowing customers to get the most out of existing options, then they'll be heading towards an iPhone-type conundrum - releasing more and more, and disappointing more and more as a result. In 2025, I believe these companies will choose one of two paths: they'll either double down on innovation at the expense of adoption, or they'll pivot to ensuring the potential of existing models is fully realised. One path will lead to stagnation; the other to sustained growth and transformation.
In 2024, we saw some major improvements in the field of generative AI: the ability to generate complex video clips, more advanced multi-modal capabilities, and the emergence of “reasoning” models (or as I like to call them, reasoning emulators).
I believe in the year ahead we will begin to see the law of diminishing returns play out in full scale regarding training bigger and bigger language models. The amount of compute power and training data to achieve even 1% bump in output quality by some arbitrary metric will compound, until it will almost make little sense to continue down this path. It seems there is an industry trend to start to think about the systems and structures to enable agent-like behaviour, using these powerful models as a foundation. More powerful LLMs will only get us so far; however, complex chains of LLM calls are how we can solve much greater problems.
As a side note, I feel there is a tangible air of “consumer AI fatigue” settling in, particularly among the younger generation. Companies are rushing to slap AI onto any feature that has keyboard input, and it seems like every second ad mentions AI in some capacity. The marketing of AI products and features will have to become much more muted to avoid the risk of consumer eyerolls.
Grok 3 and the Scale Advantage. My prediction is that in 2025, Elon Musk’s xAI Grok 3 will set a new benchmark for AI, leveraging the largest GPU cluster ever built. With 100,000 Nvidia H100 GPUs driving the Memphis Supercluster, Grok 3 is poised to surpass GPT-5 and other leading LLMs in raw power and capability, dominating the generative AI landscape—at least temporarily.
I expect there to be a rise in the number of AI-driven SaaS applications. As of 2024, we already have quite a diverse and capable selection of models from tons of companies. OpenAI, Anthropic, Microsoft and Meta were the big names just a year ago, but we’re now seeing a lot more companies releasing their own models for a variety of use cases. The question now is not will these models keep improving – they will. The real question – and the opportunity for advancement – is in how these models will be used to solve problems. We’re already seeing big businesses leverage GenAI in their day-to-day operations, and I suspect that even more medium-sized and smaller-sized businesses will start seeing the advantage. What I’m excited for is how new businesses will start thinking about AI as their foundation, from the ground-up in 2025. What fresh ideas will the pioneers of today be building out using AI, with AI as the central play? Oh, and did I forgot to mention that this is something we’re already putting into practice at Time Under Tension?
The Time Under Tension Collective
This time last year we introduced you to the Time Under Tension Collective, a group of high caliber professionals, each bringing unique perspective to the generative AI landscape and part of our growing community of AI experts. This year we’re delighted to introduce you to the growing Collective team.
Looking Back | 2024 Predictions
This time last year, we asked the Collective the same question for their 2024 predictions. With 2024 now almost behind us, we ran their predictions through Perplexity, asking for an assessment on how much of their prediction came true and from that to give each a “Manifestation rating”, the results below.
Brad | Predicted: AI will become ubiquitous and integrated into everyday products and services. Assessment: AI has become ubiquitous and integrated into everyday products and services, fully materialising this prediction.
Manifestation rating: 5/5
Ben | Predicted: Multi-modal interfaces will emerge, moving beyond traditional conversational AI. Assessment: Google Gemini and other advancements show significant progress in multi-modal interfaces. Manifestation rating: 4/5
Tim F | Predicted: Businesses will struggle with AI adoption due to choice overload and fear of obsolescence. Assessment: Many businesses are struggling with AI adoption due to choice overload and integration challenges.
Prediction rating: 4/5
Simon | Predicted: AI will enable marketing using text, images, audio, and video together. Assessment: Multimodal marketing has largely materialised with tools like Google Gemini enabling new possibilities.
Manifestation rating: 4/5
Jackson | Predicted: OpenAI's Q* algorithm will solve complex problems in physics, math, medicine, and biology. Assessment: Advancements suggest progress towards AGI, but specific claims about Q* remain speculative.
Manifestation rating: 3/5
Hayley | Predicted: AI will power health and wellness applications for personalised health insights. Assessment: AI-powered health and wellness applications are making progress, especially in wearables. Manifestation rating: 3/5
Fiona | Predicted: AI will play a significant role in design processes, creating a new role of 'AI designer'. Assessment: AI integration in design processes is growing, but the 'AI designer' role is not yet widespread.
Manifestation rating: 3/5
Josh | Predicted: Businesses will use AI-driven platforms for autonomous operations. Assessment: AI is being used more extensively in business operations, but fully autonomous AI-driven businesses are not evident.
Manifestation rating: 3/5
Jason | Predicted: Generative AI will drive a renaissance in the Metaverse. Assessment: No strong evidence of a Metaverse renaissance, but advancements in 3D generation technologies suggest some progress.
Manifestation rating: 2/5
Tim O | Predicted: AI will be used to create personality tests and personalised assistants. Assessment: Limited evidence of AI personality tests, but personalised AI assistants are gaining traction.
Manifestation rating: 2/5
