Artificial intelligence is moving beyond chatbots that generate text and images. The next stage will connect AI with scientific discovery, engineering, public services and critical infrastructure.
Europe’s proposed research programme for 2028–2034 puts this shift in view: the European Commission has listed next-generation AI among potential “moonshot” projects intended to move research towards demonstration and real-world use.
For Pakistan, the central question is whether we will help shape these technologies or rely mainly on systems designed elsewhere.
The proposal is part of Horizon Europe, the European Union’s research and innovation programme. The Commission has proposed a €175 billion budget for 2028–2034, but the proposal remains under negotiation and is not a final allocation.
Its moonshot approach is designed to connect research with deployment, drawing on European, national, public and private funding. The phrase “next-generation AI” is not yet a precise technical blueprint. It signals that Europe wants to build capability in a strategic field, alongside areas such as quantum computing, clean aviation and the space economy.
The useful lesson is that AI progress is not measured by model size alone. The next generation must be judged by whether systems are reliable, efficient, secure and useful in the settings where people need them.
It includes advances in AI for science, where algorithms help researchers analyse data and test hypotheses, and science for AI, where deeper scientific knowledge improves the tools themselves.
It also depends on less visible foundations: computing access, high-quality data, skilled teams, evaluation, safeguards and routes from laboratory research to practical use. Europe’s emerging AI-for-science effort, RAISE, reflects this emphasis on connecting compute, data, talent and research.
This broader view fits Pakistan’s needs better than an expensive race to build the largest general-purpose model. Pakistan’s National Artificial Intelligence Policy, approved in July 2025, identifies areas including education, healthcare, agriculture, energy management and climate action.
The priority now is to convert those ambitions into funded, testable programmes with clear responsibilities and outcomes. AI should address problems that matter to people and institutions, and its performance should be measured in the real conditions where it will be used.
Electricity is one such area. AI can help forecast demand and renewable generation, identify equipment faults, support maintenance planning, and coordinate storage and flexible loads.
These applications could help operators manage a changing power system, provided the tools are tested against actual network data and engineering constraints. In safety-critical settings, a confident answer is not enough: operators need systems whose limits are understood, whose outputs can be checked and whose decisions remain accountable to people.
AI infrastructure also has an energy cost. The International Energy Agency’s 2026 update estimates that data-centre electricity use could roughly double from 485 terawatt-hours in 2025 to 950 terawatt-hours in 2030, with AI-focused facilities growing faster.
Pakistan should therefore plan computing capacity alongside electricity supply, cooling, water use and network reliability. Shared, energy-efficient computing that supports universities, startups and public-interest research may deliver more value than scattered facilities with limited access or short-lived equipment.
A practical national programme should begin with shared infrastructure and governed data. Universities and research teams need affordable access to computing, technical support and secure datasets.
Data from agriculture, health, energy, transport and climate can support locally relevant research, but it must be documented and managed under clear privacy and access rules.
Urdu and regional-language resources also need sustained investment. If local languages, crops, diseases, laws and operating conditions are missing from training and evaluation, imported systems may fail to serve the people expected to use them.
Pakistan should then select a limited number of mission-driven AI programmes. Possible missions include crop and water advice for farmers, flood and heat forecasting, clinical decision support, adaptive learning in local languages, and better forecasting and maintenance for electricity networks.
Each should bring together computer scientists, sector specialists, public institutions, businesses and the communities affected. Funding should support the full path from research to field testing.
Success should be measured through outcomes such as improved reliability, access, response time, learning or cost, rather than the number of pilots launched.
People and governance are equally important. Pakistan needs researchers who can develop advanced methods, engineers who can deploy and maintain secure systems, domain specialists who can validate outputs, and public officials who can procure AI responsibly.
Training should connect to laboratories, mentors, industry placements and real projects. High-impact systems should be independently tested, their limitations disclosed, and responsibility for errors clearly assigned.
These safeguards help institutions and citizens trust systems that prove useful.
International collaboration can accelerate this work. Horizon Europe’s 2028–2034 framework is still being negotiated, and future participation rules will depend on the final programme and individual calls.
Pakistan should not assume that funding or eligibility is guaranteed. It can, however, begin building research partnerships now: identify specific strengths, connect universities and companies with European teams, mobilise diaspora researchers, and develop joint proposals in areas such as energy, climate, agriculture, health and multilingual AI.
Strong consortia take time to form, and meaningful partners arrive with expertise and a clear contribution.
Europe’s proposed moonshot is a signal that AI leadership requires more than purchasing access to software. It requires patient research, shared infrastructure, local data, skilled people, sound governance and collaboration that carries ideas into practice.
Pakistan has an approved policy and urgent challenges where these capabilities could make a difference. The next step is a focused programme that builds what the country needs, tests it carefully and expands what works.
Pakistan should not merely buy the next generation of AI; it should help create it.
This writer can be reached at habib_rahman_habib786@yahoo.com.
We welcome your contributions to The Pakistan Observer, Submit your blogs, articles, press releases, news story pitches and news features.
Please send your submissions to our News Room or Editorial team.
