AI is not without impact
Behind every model there are data centres, servers, cooling systems, networks and storage. According to the IEA, data centres consumed around 415 TWh in 2024, close to 1.5% of global electricity, and that demand could grow substantially before 2030. The impact is not abstract: it is physical infrastructure consuming physical resources.
- Energy demand is already material and growing
- Servers account for a large share of consumption
- Cooling can be very efficient or very inefficient depending on the infrastructure
- Not every use of AI carries the same cost or the same justification
Writing, generating an image and creating video do not cost the same
There is no universal figure per request: it depends on the model, the length, the resolution, the duration and the data centre processing it. But there is one clear, straightforward reality: a short text query is typically the lightest use; generating images requires considerably more computation; and video is by far the most intensive case. The problem is not thanking someone once. The problem is multiplying millions of low-value uses or requesting heavy outputs when they are not needed.
- Text: the lightest case, but not free
- Image: higher cost per computation and per generated pixel
- Video: higher cost per duration, frames and resolution
- Using AI with judgement also means avoiding unnecessary load
Impact on jobs: the real risk is a poor transition
AI does affect employment, but the most likely effect is not a uniform disappearance of roles, rather an uneven transformation of tasks and responsibilities. The ILO estimates that around 25% of global employment is in occupations potentially exposed to generative AI, with greater pressure on administrative and mechanical tasks. The damage appears when AI is introduced solely to cut costs, without training, without redesigning work and without human judgement.
- Repetitive tasks are the most exposed
- Many roles will change before they disappear
- The need for new skills will grow
- Without training, efficiency can degrade work rather than improve it
It creates new jobs
Used thoughtfully, AI can free up mechanical tasks, improve performance, open new profiles and democratise creative processes that were previously slower or more expensive. Recent OECD evidence from SMEs points in that direction: many businesses report better performance, part of the team sees reduced workload and, in most cases, the overall headcount requirement does not change. What genuinely changes are the tasks, the pace and the skills required.
- More time to decide, think and create
- New roles linked to oversight, integration and judgement
- Greater access to creative and productive tools
- More capacity without turning people into mechanical parts
What responsible use means
For us, using AI responsibly is not a cosmetic gesture. It means choosing cases more carefully, avoiding unnecessary generation, maintaining human oversight, training the team and governing use with clear criteria around risk, quality and utility. That is where sustainability stops being rhetoric and becomes practice.
- Prioritise high-value, low-waste cases
- Do not generate for the sake of generating
- Maintain review and quality criteria
- Train before scaling
- Measure real utility and usage risks
Twin Force's position
Twin Force sits neither in blind enthusiasm nor in simplistic rejection. We are interested in AI that is useful, measurable and responsible: AI that removes mechanical burden, improves processes and leaves more space for what adds the most value to human work.
- Education and awareness for teams and clients
- Responsible adoption before indiscriminate deployment
- More judgement, not just more automation
- Technology in service of human work
AISHA: AI Sustainability & Human Alignment
AISHA is an initiative created by Twin Force to raise awareness, educate and promote understanding of sustainability and AI. It starts from a simple idea: knowing what AI can do is not enough; you also need to understand what it consumes, what it transforms and the responsibility it demands. With AISHA we want to help businesses, teams and professionals use AI with more judgement, more context and more awareness.
- Clear information on energy, social and operational impact
- Training to adopt AI with more judgement and less noise
- Resources to distinguish real utility from unnecessary use
- A more mature conversation between technology, work and society
