This week, three separate pieces landed that, taken together, redraw the entire conversation about what AI actually does to a workforce.
A global marketing study finds that the companies winning with AI are not the ones adopting it the fastest, but the ones designing humans into the workflow.
A City Hall report finds that one million Londoners are in jobs highly exposed to generative AI, with women, the young, and the highly educated bearing the brunt.
And a global manufacturing essay finds that intelligent robots have already moved past pilots into Foxconn, BMW, Bosch, and Hyundai factories, quietly cutting defects, costs, and human roles at the same time.
Three reports. One pattern. AI is no longer the question. The human design around it is.
Here’s what each one says.
1. AI, alone, doesn’t deliver growth
The CMO Council’s latest research, Marketing’s Power Partners: AI and the Human Essence, produced in partnership with WongDoody, surveyed 371 senior marketing leaders globally.
The headline finding cuts cleanly through the noise: the most successful organisations are not the ones adopting AI fastest. They are the ones redesigning workflows to combine AI with human judgement, creativity, and emotional intelligence.
The CMO Council calls this group “Power Partners.” The performance gap between them and everyone else is not subtle.
- 73% of Power Partners exceed ROI expectations, against just 22% of their peers.
- 86% see moderate-to-major impact on campaign ROI, against 43% of peers.
- Nearly 70% consistently create strong emotional customer connections, against 40% of peers.
- 70% are prepared to redesign workflows for AI-human collaboration, against just 7% of peers.
- 94% have defined collaborative content processes, against 42% of peers.
The barriers holding everyone else back are stubbornly familiar: AI skills gaps, lack of trust in AI outputs, brand authenticity concerns, fragmented data, and unclear role definitions between humans and AI. None of these are technology problems. They are organisational and cultural problems.
WongDoody CEO Skyler Mattson lands the framing best:
“The real competitive edge isn’t the volume of data AI can process; it’s the human insight that turns that data into a compelling brand story.”
And the line every executive needs framed above their desk, from CMO Council CCO Tom Kaneshige’s framing of the Power Partners model:
If your marketing team’s value is defined by tasks, AI will replace it. If it’s defined by judgement, AI will amplify it.
The regional and business-model breakouts add useful texture. The US is further along on adoption but struggles to maintain emotional resonance at scale. Europe has a fragmented data problem. APAC’s challenge is cultural resistance. B2C and hybrid organisations are pulling clearly ahead of B2B, where AI is too often treated as a productivity layer rather than a workflow redesign.
2. London jobs exposure to generative AI
A new 71-page Greater London Authority report, authored by economist Jeff Dwan-O’Reilly and announced by Mayor Sir Sadiq Khan at the Bloomberg CityLab Summit in Madrid, puts the hardest numbers yet on the UK’s AI labour market debate.
The headline:
- More than 1 million Londoners are in jobs “highly or significantly exposed” to generative AI.
- More than 300,000 are in roles classified as highly exposed, mainly routine administrative work where AI can already perform a large share of daily tasks.
- A further 748,000 are significantly exposed: IT, data, secretarial, and adjacent knowledge work.
- 46% of London’s entire workforce (about 2.4 million people) sit in jobs where AI could automate part of what they do (well above the 38% UK national average).
The exposure is uneven by design.
- Women are disproportionately exposed because they are overrepresented in administrative and clerical roles.
- Young people and graduates are heavily exposed because their entry-level pathways are heavily AI-automatable.
- Lower-paid admin workers are most likely to be replaced; higher-paid professionals are most likely to benefit from productivity gains. The wealth gap implication is explicit.
The most exposed occupations named in the report: sales administrators, brokers, web designers, telephone salespeople, journalists, librarians, travel agents.
The least exposed: architects, barbers, chefs, chief executives, driving instructors, florists, undertakers, the jobs that require physical dexterity, craftsmanship, or high-stakes human judgement.
The Mayor’s framing struck a deliberate two-handed note.
Exposure does not equal job loss, AI may augment many of these roles rather than replace them. But:
“7% of large UK businesses have already used AI to cut staff. We cannot afford to be complacent.”
To act on the report, Sir Sadiq has appointed Baroness Martha Lane-Fox to chair a new London AI and Jobs Taskforce. The intervention reframes London (not the manufacturing belt) as the centre of gravity for the UK’s AI workforce conversation.
The Sunak echo is worth flagging. Last week, the former PM (now an adviser to Anthropic and Microsoft) argued that AI is “flattening the jobs market” for young people in law, accountancy, and creative industries. The GLA report quietly upgrades that argument from anecdote to dataset.
3. On Fady Jameel’s Rise of the robots
Fady Jameel’s essay turns the camera away from white-collar work and points it at the factory floor, where physical AI has already moved out of the demo phase and into measurable cost savings.
The setup: until recently, factory automation is a process centering around narrow, repetitive tasks. AI has changed that. The convergence of three capabilities is what makes the new generation of industrial robots different:
- Perception – high-resolution cameras and LiDAR allow robots to recognise objects and calculate spatial location.
- Autonomy – modern AI models like Google DeepMind’s Gemini Robotics and Nvidia’s Isaac GR00T fuse vision, language, and action to self-determine workflows.
- Manipulation – tactile sensors and ultra-soft grippers enable near-human “touch.”
The result is context-based robotics: machines that can perform tasks they were never specifically trained for. The technical term is “zero-shot learning.” The business term is “we need fewer humans.”
The numbers Jameel marshals are the most useful part of the piece:
- 4 million industrial robots deployed worldwide as of 2024, up 10% year on year.
- 70% of all newly installed robots in 2023 were in Asia, with China and Japan leading.
- Amazon: 25% cost savings and 30% more skilled in-house jobs at its Louisiana fulfilment centre after AI rollout. The “Sparrow” arm now handles 60% of warehouse inventory; “Proteus” moves alongside humans without segregated safety zones.
- Foxconn: defect rates cut by 25%, running costs down 15% using Nvidia AI plus precision-control force-feedback technology.
- BMW: vehicle defects cut by 60% via the Car2X and AIQX systems.
- Bosch: 1,500+ AI patents filed, $3 billion further investment by 2027, generative AI reducing rollout time for new AI applications from six months to a few weeks.
- ADNOC: AI-driven predictive maintenance across 2,500+ pieces of heavy equipment.
- Adidas: AI-driven personalisation lifted revenue per user by 18.5% in a single month.
- Hyundai’s Metaplant in Georgia: 850+ robots, 300 Automated Guided Vehicles, 500,000 vehicles a year, with thousands more Boston Dynamics units inbound (including Spot and Atlas).
The most quoted survey stat from the WEF and BCG: 9 in 10 companies across Asia, Europe, and the Americas plan to introduce AI to their production processes within the next three years. Early adopters average 14% savings on traditional manufacturing costs.
Jameel is honest about the constraints.
Robotics data is expensive and scarce, safety in shared spaces remains hard, and 3D spatial perception is still mechanically and sensorily limited. Synthetic data, Vision-Language-Action (VLA) models, and the new wave of “cobots” (collaborative robots like Delta’s D-Bot) are closing the gap.
His closing argument is classically optimistic: AI will free people for “more productive, rewarding and distinctly human workplace challenges.” Whether that promise gets delivered depends on the same thing every other AI report this month says it depends on: leadership, design, and policy, not the technology itself.
What the three pieces say together
Three different domains. Three different scales. One thesis.
- Marketing: the ROI gap between AI-with-humans and AI-on-its-own is now 73% versus 22%.
- London: a million workers are in roles where AI can do part of what they do; women and young people most exposed.
- Manufacturing: the global average factory is on track to see 14% cost savings as physical AI rolls out — and the human roles being absorbed are not abstractions, they are wages.
The throughline is the same shift, dressed in three different industries:
The competitive question is no longer “do you use AI?” The competitive question is “have you redesigned the work around it?”
For business leaders reading these three pieces together, three signals:
1. Workflow redesign is the actual transformation. Layering AI onto existing processes is the failure pattern across all three reports. The CMO Council data is the cleanest version: a 9× ROI gap between the redesigners and the layerers.
2. Workforce planning needs occupational mapping, not generic narratives. “AI will change jobs” is no longer a useful framing. Which roles in your organisation match the GLA’s high-exposure list? Which match the augmentation profile? The detail is now available — use it.
3. The augmentation-or-elimination question is a leadership choice. The same AI capability can lift productivity for workers or replace them. Marketing teams choosing collaboration are out-earning their peers. London’s Mayor is building a taskforce. Foxconn cut headcount and called it efficiency. The technology is neutral; the choice is not.
Three reports. One quiet message: the future of AI in business will be decided by humans deciding how to design AI into business.
