At a scientific conference in Arusha, a Tanzanian scholar’s research on AI policy drew international perspectives and government attention, opening a conversation about how prepared Tanzania is for a rapidly changing technology._
By Our Correspondent - Arusha, Tanzania
During her opening speech at the third Five University Consortium Conference in Arusha, Hon. Wanu Hafidh Ameir (MP), Deputy Minister of Education, Science and Technology, highlighted the need for a wider conversation on artificial intelligence and Tanzania’s preparedness to govern it.
She said she had noted with interest the discussion about whether AI guidelines can keep pace with the rapid growth of AI in Tanzanian higher education.
“This is precisely the kind of conversation we need, not only about what technology can do, but also about how our institutions, policies and governance frameworks should respond to its opportunities and challenges.”
The discussion she referred to had been brought into the conference by Dr Philbert Nyinondi, a lecturer and science-communication researcher at Sokoine University of Agriculture whose work sits at the intersection of science, technology, society and policy, including advocacy for the responsible and safe use of emerging technologies.
His paper, “Can AI Guidelines Keep Pace with Artificial Intelligence? Assessing Policy Adaptability in Tanzanian Higher Education,” examined Tanzania’s two 2025 national guidelines relevant to artificial intelligence and digital education in universities.
Nyinondi was not asking whether Tanzania has AI policy. His question was more demanding: whether the policy system can adapt quickly enough as the technology changes.
The study found that Tanzania already has important foundations for adaptive AI governance. The guidelines provide for monitoring and evaluation, institutional responsibilities, stakeholder feedback, evidence generation, policy review and continuous improvement. The more difficult gap concerns the mechanisms for responding rapidly when a significant AI capability, opportunity or risk emerges.
The question gained wider meaning at the third Five University Consortium Conference, hosted by the Nelson Mandela African Institution of Science and Technology in Arusha from 23 to 25 September. The meeting brought together researchers and scholars around the theme “From Dialogue to Impact: Integrating Science, Technology, Policy and Climate Resilience for a Sustainable Future.”
What followed was a conversation that moved from the relationship between technology and society, to the realities of building AI for the Global South, and finally to the institutions and policies needed to govern it.
When technology and society move apart
Prof Raffaella Cagliano of Politecnico di Milano, Italy, approached the future of technology from the perspective of society.
Her concern was that technological development and human needs can move in different directions. Some technologies are developed without a sufficiently clear understanding of the problems they are intended to address, while genuine social needs may remain without appropriate technological solutions.
“Technology and needs often don't really talk to each other,” Cagliano told the conference.
She argued for a socio-technical approach in which technological development and the social system are considered together.
“We cannot develop technology in a separate way compared with the social variable of the system.”
For the AI debate, the implication is significant. The question cannot simply be what artificial intelligence is capable of doing. It must also consider where it is being used, whose needs it serves and the social conditions in which it operates.
Making AI work in the real world
Prof Tushar Shinde, an Indian scholar working at IIT Madras Zanzibar, approached the challenge from the technological side.
His focus was efficient AI and the difficulty of making increasingly powerful systems usable in environments where computing resources, electricity, connectivity and data may be limited.
Shinde discussed approaches including model compression, pruning, quantisation and data-efficient training, challenging the assumption that bigger AI models necessarily represent greater progress.
“The goal is not to build a bigger model. However, the goal is how to make the model smaller, faster and cheaper.”
For countries in the Global South, the argument has practical consequences. AI that demands enormous computing resources may be powerful, but its development and use can remain beyond the reach of institutions and communities with limited infrastructure.
Shinde therefore proposed impact as a more meaningful measure of progress.
“The future of AI should not be measured by the model size, but how much impact it is making.”
His concluding message connected efficiency to the pace of AI development: “As AI is moving fast, let us not only build AI faster, let us build AI that goes farther.”
The two international perspectives widened the question that Nyinondi had brought into the room. If technology must respond to society, and AI must work within real social and resource conditions, the policies governing it must also be capable of responding to change.
The policy gap is about speed
Nyinondi’s analysis compared the National Guidelines for Artificial Intelligence in Education (NGAIE, 2025) with the National Digital Education Guidelines for Universities (NDEGU, 2025).
The findings point to a policy framework that already recognises adaptation. Both guidelines contain provisions on monitoring, institutional responsibility, stakeholder participation, evidence, continuous improvement and policy review.
The unresolved issue is what happens between scheduled reviews.
A new AI capability may emerge rapidly. A new risk may become visible through university experience or research. An application may become widespread before policymakers have had time to assess its implications.
The study found that the guidelines do not clearly establish a technology-triggered threshold for an urgent policy review. Dedicated horizon-scanning and early-warning mechanisms are also less clearly specified, as is the complete pathway from detection and reporting through escalation, decision-making and formal AI-specific policy adjustment.
Nyinondi describes this as the difference between adaptive policy intent and operational policy agility.
A policy can recognise that technology will change. A more agile policy system needs to establish how that change is detected, who acts on it and how quickly evidence can become a decision.
That distinction becomes increasingly important as AI moves into teaching, assessment, research, academic administration and other areas of university life.
From a conference paper to a national conversation
The Deputy Minister’s response gave the research a significance beyond its original focus on higher education.
By highlighting the need for a conversation about AI policy and preparedness, Hon. Wanu Hafidh Ameir (MP) placed the issue within the broader responsibility of government, universities and other institutions to prepare for emerging technologies.
Her intervention also echoed a central purpose of the conference: moving research beyond academic discussion towards policy, practice and development impact.
The significance of the discussion lies in this widening circle. What began with two higher-education policy documents now raises questions about institutional readiness, regulatory responsiveness and the country’s ability to learn from technological change.
Part of a bigger STI conversation
The timing is also important for Tanzania’s wider science, technology and innovation agenda.
In 2026, the Tanzania Commission for Science and Technology (COSTECH) is marking 40 years, with COSTECH@40 providing a platform to reflect on four decades of science, technology and innovation and consider the country’s direction towards Vision 2050. The anniversary programme places emerging technologies and digital transformation within a broader national reflection on how science, technology and innovation can contribute to development.
The AI discussion emerging from Arusha fits naturally within that reflection. The issue is not simply how Tanzania adopts AI, but whether its research institutions, universities, regulators and policymakers can develop the capacity to understand technological change and respond to it in time.
That requires a conversation extending beyond government. Researchers need to identify emerging opportunities and risks. Universities need to generate evidence from actual use. Technology developers need to engage with local realities. Policymakers need mechanisms through which evidence can inform timely decisions.
The conversation beyond Arusha
The three academic perspectives offered different starting points but converged around a common concern.
Cagliano placed human need at the centre of technological development. Shinde challenged researchers to make AI efficient and accessible within real-world constraints. Nyinondi brought the question into policy, asking whether governance can adapt as the technology changes.
The Deputy Minister then placed the issue within Tanzania’s national policy conversation.
The result is a question that reaches beyond higher education and beyond the conference itself: how can Tanzania ensure that its capacity to govern emerging technologies develops alongside its capacity to create and use them?
That question fits squarely within the wider STI reflection taking place through COSTECH@40 and Tanzania’s longer-term Vision 2050 ambitions.
AI will continue to evolve. New applications, opportunities and risks will emerge between policy cycles. Tanzania’s challenge is to ensure that its institutions can recognise those changes, learn from evidence and respond with the speed and accountability that the technology demands.
The conversation has started in Arusha. Its real test will be what happens after the conference walls disappear.
She said she had noted with interest the discussion about whether AI guidelines can keep pace with the rapid growth of AI in Tanzanian higher education.
“This is precisely the kind of conversation we need, not only about what technology can do, but also about how our institutions, policies and governance frameworks should respond to its opportunities and challenges.”
The discussion she referred to had been brought into the conference by Dr Philbert Nyinondi, a lecturer and science-communication researcher at Sokoine University of Agriculture whose work sits at the intersection of science, technology, society and policy, including advocacy for the responsible and safe use of emerging technologies.
His paper, “Can AI Guidelines Keep Pace with Artificial Intelligence? Assessing Policy Adaptability in Tanzanian Higher Education,” examined Tanzania’s two 2025 national guidelines relevant to artificial intelligence and digital education in universities.
Nyinondi was not asking whether Tanzania has AI policy. His question was more demanding: whether the policy system can adapt quickly enough as the technology changes.
The study found that Tanzania already has important foundations for adaptive AI governance. The guidelines provide for monitoring and evaluation, institutional responsibilities, stakeholder feedback, evidence generation, policy review and continuous improvement. The more difficult gap concerns the mechanisms for responding rapidly when a significant AI capability, opportunity or risk emerges.
The question gained wider meaning at the third Five University Consortium Conference, hosted by the Nelson Mandela African Institution of Science and Technology in Arusha from 23 to 25 September. The meeting brought together researchers and scholars around the theme “From Dialogue to Impact: Integrating Science, Technology, Policy and Climate Resilience for a Sustainable Future.”
What followed was a conversation that moved from the relationship between technology and society, to the realities of building AI for the Global South, and finally to the institutions and policies needed to govern it.
When technology and society move apart
Prof Raffaella Cagliano of Politecnico di Milano, Italy, approached the future of technology from the perspective of society.
Her concern was that technological development and human needs can move in different directions. Some technologies are developed without a sufficiently clear understanding of the problems they are intended to address, while genuine social needs may remain without appropriate technological solutions.
“Technology and needs often don't really talk to each other,” Cagliano told the conference.
She argued for a socio-technical approach in which technological development and the social system are considered together.
“We cannot develop technology in a separate way compared with the social variable of the system.”
For the AI debate, the implication is significant. The question cannot simply be what artificial intelligence is capable of doing. It must also consider where it is being used, whose needs it serves and the social conditions in which it operates.
Making AI work in the real world
Prof Tushar Shinde, an Indian scholar working at IIT Madras Zanzibar, approached the challenge from the technological side.
His focus was efficient AI and the difficulty of making increasingly powerful systems usable in environments where computing resources, electricity, connectivity and data may be limited.
Shinde discussed approaches including model compression, pruning, quantisation and data-efficient training, challenging the assumption that bigger AI models necessarily represent greater progress.
“The goal is not to build a bigger model. However, the goal is how to make the model smaller, faster and cheaper.”
For countries in the Global South, the argument has practical consequences. AI that demands enormous computing resources may be powerful, but its development and use can remain beyond the reach of institutions and communities with limited infrastructure.
Shinde therefore proposed impact as a more meaningful measure of progress.
“The future of AI should not be measured by the model size, but how much impact it is making.”
His concluding message connected efficiency to the pace of AI development: “As AI is moving fast, let us not only build AI faster, let us build AI that goes farther.”
The two international perspectives widened the question that Nyinondi had brought into the room. If technology must respond to society, and AI must work within real social and resource conditions, the policies governing it must also be capable of responding to change.
The policy gap is about speed
Nyinondi’s analysis compared the National Guidelines for Artificial Intelligence in Education (NGAIE, 2025) with the National Digital Education Guidelines for Universities (NDEGU, 2025).
The findings point to a policy framework that already recognises adaptation. Both guidelines contain provisions on monitoring, institutional responsibility, stakeholder participation, evidence, continuous improvement and policy review.
The unresolved issue is what happens between scheduled reviews.
A new AI capability may emerge rapidly. A new risk may become visible through university experience or research. An application may become widespread before policymakers have had time to assess its implications.
The study found that the guidelines do not clearly establish a technology-triggered threshold for an urgent policy review. Dedicated horizon-scanning and early-warning mechanisms are also less clearly specified, as is the complete pathway from detection and reporting through escalation, decision-making and formal AI-specific policy adjustment.
Nyinondi describes this as the difference between adaptive policy intent and operational policy agility.
A policy can recognise that technology will change. A more agile policy system needs to establish how that change is detected, who acts on it and how quickly evidence can become a decision.
That distinction becomes increasingly important as AI moves into teaching, assessment, research, academic administration and other areas of university life.
From a conference paper to a national conversation
The Deputy Minister’s response gave the research a significance beyond its original focus on higher education.
By highlighting the need for a conversation about AI policy and preparedness, Hon. Wanu Hafidh Ameir (MP) placed the issue within the broader responsibility of government, universities and other institutions to prepare for emerging technologies.
Her intervention also echoed a central purpose of the conference: moving research beyond academic discussion towards policy, practice and development impact.
The significance of the discussion lies in this widening circle. What began with two higher-education policy documents now raises questions about institutional readiness, regulatory responsiveness and the country’s ability to learn from technological change.
Part of a bigger STI conversation
The timing is also important for Tanzania’s wider science, technology and innovation agenda.
In 2026, the Tanzania Commission for Science and Technology (COSTECH) is marking 40 years, with COSTECH@40 providing a platform to reflect on four decades of science, technology and innovation and consider the country’s direction towards Vision 2050. The anniversary programme places emerging technologies and digital transformation within a broader national reflection on how science, technology and innovation can contribute to development.
The AI discussion emerging from Arusha fits naturally within that reflection. The issue is not simply how Tanzania adopts AI, but whether its research institutions, universities, regulators and policymakers can develop the capacity to understand technological change and respond to it in time.
That requires a conversation extending beyond government. Researchers need to identify emerging opportunities and risks. Universities need to generate evidence from actual use. Technology developers need to engage with local realities. Policymakers need mechanisms through which evidence can inform timely decisions.
The conversation beyond Arusha
The three academic perspectives offered different starting points but converged around a common concern.
Cagliano placed human need at the centre of technological development. Shinde challenged researchers to make AI efficient and accessible within real-world constraints. Nyinondi brought the question into policy, asking whether governance can adapt as the technology changes.
The Deputy Minister then placed the issue within Tanzania’s national policy conversation.
The result is a question that reaches beyond higher education and beyond the conference itself: how can Tanzania ensure that its capacity to govern emerging technologies develops alongside its capacity to create and use them?
That question fits squarely within the wider STI reflection taking place through COSTECH@40 and Tanzania’s longer-term Vision 2050 ambitions.
AI will continue to evolve. New applications, opportunities and risks will emerge between policy cycles. Tanzania’s challenge is to ensure that its institutions can recognise those changes, learn from evidence and respond with the speed and accountability that the technology demands.
The conversation has started in Arusha. Its real test will be what happens after the conference walls disappear.
No comments:
Post a Comment