AI is no longer the variable. How delivery organisations adapt around it will decide what breaks next. By now, it should be clear that the debate about whether AI belongs in project delivery is over. It does. It is already embedded in how analysis is produced, how options are evaluated, and how decisions move through delivery environments. In many organisations, it is shaping outcomes more quietly and pervasively than any previous digital tool. The more difficult conversation

A single line of AI output exposed gaps in safety, accountability and professional judgement. In late 2024, Google’s Gemini AI chatbot delivered a threatening response to a student during a routine homework interaction, ending with the words “Please die.” The incident was widely reported and acknowledged by Google as a failure of its safety systems. While the event shocked the public, its implications run far deeper for organisations embedding generative AI into delivery env

Large language models produce fluent answers, but their inability to handle numbers effectively is becoming a material risk to project delivery. Large language models are increasingly embedded in project environments, supporting reporting, planning and decision preparation. However, research highlighted by BoundaryML points to a persistent and structural limitation: LLMs do not understand numbers in the way project delivery requires. They do not calculate, verify or reason ov

As generative tools become easier to tune and personalise, the real risk is no longer technical failure but misplaced certainty. Enthusiasm has always accompanied the arrival of new tools in project delivery. Early adopters experiment, promising use cases circulate, and confidence grows faster than evidence. In most technology cycles, this phase resolves itself as limitations become clearer and practice matures. AI is unfolding differently. Over the past year, generative too

AI did not arrive as a disruptor. It came as a stress test, and many delivery models quietly failed it. It is tempting to describe 2025 as the year artificial intelligence finally entered project delivery in a meaningful way. Autonomous agents became usable. Predictive systems improved. Automation moved closer to the core of delivery work rather than hovering at the edges. That description is not wrong, but it misses the more important story. AI did not so much transform proj

What happens when the shiniest new technology stops dazzling us and starts demanding something far more difficult: actual change? John Hetherington brought the rain but not the cold when he flew in from Canada to London, a fitting metaphor for the conversation that followed. Because whilst 2025 didn't freeze AI innovation in its tracks, it certainly dampened some of the euphoria that characterised the previous three years. And according to John, that's precisely what needed

OpenAI’s latest Codex model is not about writing better code faster. It is about compressing delivery cycles and redefining how engineering work is organised. There is a tendency to treat every new AI model release as another incremental step forward. In the case of GPT-5.2-Codex , that framing would undersell what is actually changing. Announced by OpenAI in December 2025, GPT-5.2-Codex is positioned as the most capable code-focused model OpenAI has released to date. It buil

OpenAI’s latest image generation model rollout signals a strategic pivot as competition intensifies and the company mobilises in “Code Red” mode to protect leadership in AI. A Strategic Sprint, Not Just a Product Update In December 2025 , TechCrunch  reported that OpenAI had rolled out a significant update to ChatGPT’s image generation capabilities amid what has been described internally as a “Code Red” push. The upgraded model rolling out as GPT Image 1.5 brings significant

Surveys show strong confidence in artificial intelligence across construction. The real test will be whether optimism can survive data constraints, fragmented delivery models and commercial pressure. Construction is not usually the first industry to embrace technological optimism . It is capital-intensive, risk-averse and shaped by contracts that reward certainty over experimentation. That is precisely why recent findings on contractor attitudes to artificial intelligence des

A proposed class-action lawsuit against Adobe claims that unauthorised books were used to train its AI models. The legal and operational consequences extend far beyond a single company. The Lawsuit That Could Redefine AI Training Norms In December 2025 , Adobe Inc. was hit with a proposed class-action lawsuit in the United States alleging that the company used copyrighted books without permission to train one of its artificial intelligence models. The case was filed in the U

McKinsey’s decision to cut thousands of roles as AI reshapes its operating model signals a structural shift in how professional services value work, talent and delivery. Why This Decision Matters More Than It First Appears When The Times  reported that McKinsey & Company is preparing to eliminate thousands of roles as artificial intelligence advances , the headline framed it as another chapter in the consulting sector slowdown. That reading misses the point. This is not simpl

Imagine walking into a bar and hearing your favourite song. The production is flawless, the arrangement ideally suited to the venue's atmosphere. You close your eyes and let the melody wash over you. Then someone leans across and says, "That's AI." Your eyes snap open. Does it feel different now? More hollow? The same, but somehow less real? This is the dilemma facing the music industry in 2025, and it's far more nuanced and troubling than most of us realise. Two accomplished

A new government standard for project and programme data might be the most foundational breakthrough of 2025, but it matters more than most AI announcements. On 11 December 2025, while everyone was distracted by OpenAI's latest model release and Trump's AI executive order, the UK government quietly launched something that will actually improve project delivery: a new standard for programme and project data. No flashy demo. No viral moment. Just a methodical attempt to solve

Google's partnership with the UK government focuses on the unglamorous work of identifying new materials that could transform infrastructure delivery. The headlines about AI tend to focus on chatbots, image generators, and automation replacing white-collar jobs. Meanwhile, Google DeepMind just announced something far more consequential: a fully automated materials science laboratory in the UK that will use AI and robotics to synthesise and test hundreds of materials per day.

The Mouse House's OpenAI partnership signals Hollywood's surrender to generative AI, but will it increase creativity or just flood the internet with corporate slop? Walt Disney spent decades building an empire on the premise that quality animation required armies of skilled artists labouring over every frame. On 11 December 2025, his company paid $1 billion for the right to let anyone generate videos of Mickey Mouse, Darth Vader, and Elsa using artificial intelligence. No art

The attempt to override state regulations might backfire spectacularly for project delivery leaders navigating AI governance. On 11 December 2025 , President Trump signed an executive order that Silicon Valley had been lobbying for and state regulators had been dreading. Flanked by AI advisor David Sacks, Commerce Secretary Howard Lutnick, and Senator Ted Cruz in the Oval Office, Trump declared there would be only "one rulebook" for AI regulation in America. No more navigatin

A modest improvement rushed to market proves the AI arms race matters more than breakthrough innovation. The corridors of OpenAI echoed with urgency in early December. Sam Altman's internal 'code red' memo had landed like a bombshell just days earlier, warning staff that ChatGPT traffic was declining and Google's Gemini 3 was eating their lunch. Some employees reportedly pushed back, arguing GPT-5.2 needed more time in the oven. They were overruled. On 11 December 2025, Open

A conversation with Doug Hubbard and Andreas Leed on measurement, risk, and the decisions that determine project success. The Statistic Nobody Wants to Discuss Only one in every two hundred projects finishes on time, within budget, and delivers the benefits outlined in the original business case . Yet the project management profession essentially treats this figure as inevitable. Rather than questioning the fundamental assumptions that produce this outcome, organisations acce

A Mathematical Breakthrough, a Major Caveat, and What It Tells Us About AI's Real Trajectory. An AI system called Aristotle, built by Harmonic (founded by Robinhood CEO Vlad Tenev), independently solved a mathematical problem that has remained open for nearly 30 years. Erdős Problem # 124, posed in 1995, is the sort of mathematical challenge that separates the brilliant from the merely expert. The AI didn't just solve it. It solved it in six hours, then formally verified the

When Market Leadership Becomes a Fragile Thing: Why Sam Altman's emergency memo signals deeper troubles in the AI industry's viability. Sam Altman declared a "code red" emergency at OpenAI on Monday, and the tech industry spent the rest of the week parsing its meaning. The CEO's internal memo to staff was explicit: redirect resources toward improving ChatGPT immediately. Delay other projects. This is the priority. The immediacy surprised some observers. After all, OpenAI rema