2025 AI Predictions: Transformative Tech, Smarter Workflows, and Industry Insights

This Week’s BIG thing

Our predictions for AI in 2025

In 2025, AI is going to get even smarter and more efficient! With impressive models like OpenAI’s O3 and a range of affordable, specialised tools, we can look forward to exciting improvements in memory, multimodal workflows, and safety protocols. These advancements are set to transform the way we work. Here's a glimpse of what we think is coming up in the next year.

1. Larger AI Models

In 2024, the intelligence of AI advanced significantly through scaling network size, training data, and training duration. As we move into 2025, this scaling will remain a critical factor in developing more intelligent models. However, challenges like ‘peak data’—the scarcity of high-quality training data—may start to limit progress. OpenAI’s O3 model suggests a shift towards enhancing reasoning capabilities by extending processing times and improving alignment with human values.

2. Smaller, Lite AI Models

As the operational costs of large AI models became more apparent in 2024, businesses began to prioritise smaller, specialised models. These cost-effective solutions, such as GPT-4o-mini, Meta’s Llama, and Google’s Gemini 2.0 Flash, are expected to become standard in organisations by 2025. While advanced AGI remains a priority for some, many companies will find greater value in these more affordable and scalable alternatives. Moore’s law may also reduce the cost barrier to accessing top-tier AI by 2026.

3. Agentic Workflows with Multimodal Architectures

AI agents are evolving towards more sophisticated cognitive architectures. Progress will be seen in the transition from single-agent models to multi-agent systems defined by humans and then to autonomous agentic workflows with human oversight. By 2025, these systems will increasingly incorporate multimodal inputs such as images, videos, and PDFs. Some organisations may even deploy self-governing multi-agent systems by 2026, although widespread practical application will likely be limited due to accuracy and reliability concerns.

4. Significant Improvements to Memory and Attention

Anticipated advancements like Google’s infini-attention and Microsoft’s near-infinite memory will transform AI’s ability to handle complex tasks. These improvements will enhance scalability, recall, reasoning, and accuracy, allowing models to process larger volumes of data, such as entire books or extensive project documentation, without the need for segmenting inputs. Faster, more reliable interactions will favour one-shot prompting, making AI an even more powerful tool for professionals.

5. Robotic AI

While the idea of AI-powered robots may seem futuristic, 2025 will see their increasing integration into industries like manufacturing and construction, enhancing efficiency in the background. Human-like robots, as showcased by Tesla, remain a distant reality, but more advanced robotics will quietly improve operations. In healthcare, autonomous robots will address labour shortages by providing caregiving solutions and improving patient outcomes, while synthetic data will aid in the safer training of healthcare AI systems.

6. Generative AI’s Unprecedented Impact

Generative AI adoption has already outpaced earlier technological revolutions, driving productivity and creativity across industries. However, organisations may face slower-than-expected adoption due to the need for cautious, well-informed strategies. Leaders will focus on maximising value by integrating generative AI in a way that supports operational success.

7. AI-Driven Wealth Management

AI copilots are revolutionising financial advisory, delivering hyper-personalised insights, enhancing efficiency, and streamlining client engagement. Although reasoning models remain a work in progress, these tools are already making financial services more accessible and tailored to individual needs.

8. Healthcare Robotics and RNA Therapeutics

Autonomous robotics will continue to address critical healthcare labour shortages, improving patient care and outcomes. Meanwhile, RNA technologies, including mRNA vaccines, will drive breakthroughs in treating previously untreatable diseases. These advancements highlight the importance of small, finely-tuned data models for achieving precise and cost-effective solutions.

9. The Future of Data Centres

The growing demand for AI workloads is reshaping the design and functionality of data centres, with over £1 trillion in investments expected. However, the increasing complexity of AI pipelines will emphasise the need for better monitoring and quality management to avoid costly errors.

10. Spatial Computing’s Enterprise Revolution

Spatial computing devices, such as the Apple Vision Pro, are poised to transform workflows and collaboration across sectors. These technologies will benefit from advances in unstructured data analysis, enabling enterprises to harness their data assets in innovative and immersive ways.

11. AI Mergers & Acquisitions Surge

The acceleration of AI mergers and acquisitions will see companies integrating cutting-edge technologies to stay ahead. However, the practical deployment of agentic AI remains challenging, particularly in managing multi-step workflows and ensuring reliability.

12. Crypto Meets AI Payments

Cryptocurrency will emerge as a seamless payment method for AI agents, bypassing traditional banking systems. This convergence of AI and crypto will be supported by the blending of analytical and engineering roles within data teams, enabling advanced use cases.

13. AI Leadership: US vs. China

The competition between the US and China for AI supremacy will intensify, shaping the global innovation landscape. As this race continues, businesses will need to address the complexity of expanding pipelines and ensure data quality to maintain a competitive edge.

14. ASL-3 on AI Safety

As AI systems become more intelligent and users more adept, the potential for misuse or unintended behaviours grows. By 2025, we may reach Anthropic’s AI Safety Level 3 (ASL-3), which will heighten the risks of misuse in sensitive fields such as chemistry, biology, and nuclear science. This will require robust safeguards to mitigate societal risks.

15. Revolutionising Retail with AI

AI-driven personalisation will reshape the shopping experience, boosting customer satisfaction and conversions. The focus will shift towards deploying robust solutions that prioritise results over mere tooling, enabling retailers to create meaningful and measurable impacts.

Other Predictions

Our other predictions include enhanced precision and interpretability of existing models like GPT-4o and Claude Sonnet 3.5, and a shift towards more anthropomorphic interactions with emerging, more powerful AI systems such as o1. We’ll cover these in more detail in 2025!

What’s new: Tech

Generative AI Adoption Soars, Outpacing PCs and the Internet

Introduction

Generative AI is rewriting the history of technology adoption, achieving widespread use at a pace twice as fast as personal computers and the internet, according to a recent study from the National Bureau of Economic Research (source). This breakthrough technology is not only redefining industries but also transforming how work gets done, with profound implications for construction, energy, and real estate sectors.

What Does This Mean for Me?

This generative AI boom isn’t just a tech trend—it’s a game-changer for productivity and project outcomes.

• Enhanced Efficiency: AI-powered tools are streamlining processes like design, planning, and risk analysis, cutting down time and errors.

• Data-Driven Decision-Making: Real-time insights from generative AI can optimise resource allocation, reduce costs, and improve sustainability in project delivery.

• Skills Shift: With the rapid adoption of AI, there’s a growing need for upskilling in digital literacy and AI fluency to stay competitive in project-driven industries.

• Competitive Edge: Early adopters of generative AI solutions are likely to see significant advantages in innovation, collaboration, and overall project success.

Key Themes or Details

• Rapid Adoption: Generative AI reached 1 million users in just 5 days, a milestone PCs took years to achieve.

• Industry Disruption: Sectors from construction to real estate are leveraging AI to automate repetitive tasks, boost creativity, and enhance project planning.

• Economic Impact: Companies adopting generative AI report notable cost savings and productivity gains, further accelerating its integration into workflows.

• Barriers to Entry Lowered: As tools like ChatGPT and DALL·E become more accessible, businesses of all sizes can harness AI’s potential.

• Collaboration Benefits: Generative AI fosters improved teamwork through better communication, visualisation, and real-time project adjustments.

Down the Rabbit Hole

_{• The NBER Study on Generative AI Adoption}

💙 Other tech news we’re loving

• Siri Meets ChatGPT: In iOS 18.2, Siri now directs queries to ChatGPT, unlocking new AI capabilities for professionals. Learn how it works.

• GitHub Copilot Goes Free: Microsoft makes GitHub Copilot free, empowering developers to integrate AI into workflows effortlessly. Explore the benefits.

• Elon Musk’s xAI Secures $6 Billion: xAI’s latest funding propels its ambitious AI projects, signalling significant industry shifts. Dive into the details.

• Trump’s AI Policy Picks: The new administration’s tech team outlines its vision for AI and digital innovation. Read the story.

• Waymo Outperforms Humans in Safety: Waymo’s self-driving tech excels in preventing injuries and damage, setting new standards for automation. See the data.

• ChatGPT Reads Apple Notes: OpenAI’s latest update integrates Apple Notes, streamlining knowledge workflows for Mac users. Discover more.

•Databricks Raises $10 Billion: Databricks secures major funding, reinforcing its leadership in data-driven AI solutions. Unpack the implications.

What’s new: Projects

Encyclopedia Britannica Reinvents Itself as an AI Company

The iconic Encyclopedia Britannica, synonymous with reliable knowledge for over two centuries, has taken a bold leap into the future by rebranding itself as an AI company (source). Leveraging cutting-edge technology, Britannica now aims to enhance knowledge accessibility and create dynamic, AI-driven educational tools tailored for the modern age.

What Does This Mean for Me?

This transformation means smarter, AI-powered resources are on the horizon, poised to reshape how critical knowledge is accessed and applied.

• Streamlined Learning: AI-driven tools will make it easier to find and understand complex topics relevant to project management and technical fields.

• Bespoke Knowledge Delivery: Industry-specific content, powered by Britannica’s AI, could offer real-time, actionable insights tailored to professionals.

• Improved Training Programs: Interactive and dynamic AI tools may revolutionise workforce education, making upskilling and training more effective.

• Strategic Decision Support: Enhanced access to curated information will empower leaders to make data-informed decisions quickly and confidently.

Key Themes or Details

• Historic Pivot: Britannica’s shift to AI underscores a broader trend of established institutions embracing generative technologies.

• Enhanced Tools: New AI-driven solutions will provide smarter search functionalities, interactive learning experiences, and personalized content recommendations.

• Legacy Meets Innovation: The company blends its longstanding reputation for accuracy with the latest in machine learning and natural language processing.

• Broad Applications: Beyond education, Britannica’s AI solutions may expand into areas like industry-specific knowledge bases and decision-support tools.

• Future-Proofing: This move aligns Britannica with the growing demand for reliable, AI-driven content in professional and personal contexts.

Down the Rabbit Hole

_{• Read about Britannica’s transformation}

As Encyclopedia Britannica enters the AI era, its reinvention reflects the growing synergy between knowledge and technology—bringing professionals in all sectors a step closer to smarter, more efficient ways of working.

💙 Other Project news we’re loving

 • Saudi Arabia Bets on Robots for Neom: Robots take center stage in Neom’s ambitious construction, pushing boundaries in automated project delivery. Discover more about this futuristic initiative.

• New Rules for Autonomous Vehicles: The NHTSA introduces a regulatory framework to ensure safety in autonomous vehicle development and deployment. Explore the implications.

• Microsoft Expands AI in Healthcare: Microsoft accelerates its AI health initiatives, blending innovation with patient care advancements. Learn how it’s reshaping the health sector.

• Ireland’s 3D-Printed Social Housing: A groundbreaking project brings affordable, sustainable housing to Ireland using 3D printing. Read about this transformative initiative.

• AI Exoskeletons Revolutionizing Sites: AI-powered exoskeletons show promise in reducing back injuries on construction sites. See how this tech is advancing worker safety.

•Is Construction Ready for AI?: The sector grapples with AI adoption challenges, balancing potential with readiness. Dive into the debate.

What’s new: Productivity

Why LLMs Will Always Hallucinate: A Deep Dive into AI Limitations

Large Language Models (LLMs) like ChatGPT, Claude and Gemini are powerful tools transforming industries, but a recent study (source) argues that hallucination—producing incorrect or fabricated information—is an inherent limitation of these systems. The paper explores the mathematical and practical reasons why LLMs will always struggle with perfect factual accuracy, raising important implications for professionals relying on AI in their workflows.

What Does This Mean for Me?

For professionals, where accuracy and reliability are paramount, this insight is a crucial reminder of AI’s limitations.

• Fact-Checking is Essential: AI can streamline tasks but must be paired with rigorous validation processes to avoid costly mistakes.

• Risk Mitigation: Understanding the boundaries of AI accuracy helps mitigate the risk of hallucinations affecting decisions, plans, or reports.

• Tool Selection Matters: Incorporating domain-specific AI tools, trained on curated datasets, can reduce but not eliminate the risk of hallucination.

• Upskilling Opportunities: Awareness of AI’s limitations can help professionals fine-tune prompts and develop workflows that extract the best results while minimising errors.

Key Themes or Details

• Mathematical Basis for Hallucination: The study explains how probabilistic models, by design, prioritize plausibility over verifiable accuracy.

• Imperfect Training Data: LLMs learn from vast, imperfect datasets, making hallucination an inevitable byproduct of incomplete or biased information.

• Balancing Creativity and Accuracy: Models designed to generate creative outputs often sacrifice factual correctness, highlighting the trade-off in generative AI.

• Improvement vs. Perfection: Continuous advancements will reduce but never entirely eliminate hallucination, emphasising the need for human oversight.

• Sector Impacts: In industries like construction and real estate, even small inaccuracies can cascade into significant financial or operational risks, making cautious implementation critical.

Down the Rabbit Hole

_{• The Full Paper on Arxiv}

As LLMs continue to integrate into workflows, understanding their strengths and limitations ensures they remain valuable tools rather than potential liabilities. Stay informed, stay critical, and keep leveraging AI with care.

💙 Other productivity news we’re loving

• ChatGPT Now Understands Real-Time Video: OpenAI’s ChatGPT adds real-time video comprehension, expanding its potential for interactive, visual tasks in project delivery. Learn how this innovation transforms AI capabilities.

Project Flux Podcast

Project Flux Boxing Day Special: Unveiling OpenAI’s O3 Model

In this festive episode of Project Flux, James Garner and Yoshi Soornack unpack OpenAI’s groundbreaking O3 model, exploring its AGI potential, safety protocols, and emotional intelligence challenges. They also dive into its implications for project management and society. Listen here for cutting-edge AI insights!

One More Thing

As we bid farewell to an incredible year, we want to extend our warmest wishes for a very happy and prosperous New Year to all our readers! We are thrilled to embark on this exciting journey into the future, where the power of AI continues to revolutionise the way we deliver projects. With each passing day, AI is becoming an even more integral part of our toolkit, helping us to innovate, streamline processes, and achieve unprecedented levels of efficiency and success. We can't wait to ride this wave of technological advancement with you, exploring new horizons and unlocking endless possibilities. Here's to a year filled with groundbreaking achievements, collaboration, and the joy of seeing our projects come to life in ways we never imagined possible. Cheers to a fantastic New Year ahead!

Meet Project Flux: About Us

At Project Flux, we're committed to pioneering the future of construction and project delivery through the lens of cutting-edge Artificial Intelligence insights. Our vision is to be at the forefront of integrating AI into the fabric of project delivery, transforming how projects are conceptualised, planned, and executed.

Our Mission Project Flux aims to not only inform and educate but also to inspire professionals in the construction industry to embrace the transformative potential of AI. We believe in the power of AI to revolutionise project delivery, making it more efficient, predictive, and adaptable to the dynamic demands of the modern world.

What We Offer Through our insightful newsletters, podcasts and curated content on LinkedIn, and engaging discussions, Project Flux serves as a resource for professionals seeking to stay ahead in their field. We offer a blend of practical advice, thought leadership, and the latest developments in AI and construction technology.

What People Say…