Ahya

One recurring perpective from the Middle East, North Africa and Pakistan is the lack of knowledge and capacity (technical and legislative) to build effective policy frameworks for AI overnance. Many countries in the MENAP region, donot even have basic data protection laws, and rights. Hence rapid AI adoption, is a larger concern as policy makers are unable to effectively legislate. Wherever, legislation has been attempted the supporting implementation mechanism is absent. Capacity is also an issue when it comes to developing countries across the region, and their Ministry's of IT, very few comptent professionals exist. The political systems mostly based on tribal Additionally, developing countries also lack the ability to develop sovereign AI capabilities. While many policy-makers, make commitments, the development finance required to underwrite such capabilities does not exist. The exisiting DFI's that normally finance development in our region, are mainly geared to assess infrastructure, banking (or services sectors excluding technology). The UN-ITU, must raise this as an urgent financing gap that needs to be met to develop sovereign AI capabilities for MENAP countries in order to build effective governance models for regulating the same. I can further add to this as having been a Board Member of IGNITE, the National Technology Fund of Pakistan and having worked on UAE's reguatory frameoworks for technology, AI and sustainability.

• Open-source software, open data and open AI models
• AI capacity-building
• Transparency, accountability, and human oversight
• Social, economic, ethical, cultural, linguistic and technical implications of AI

For the MENAP region and developing regions in particular, AI capacity building and understanding of the technology is the biggest fundamental challenge. For developing countries across the region, a lack of underpinning regulation on data protection is the missing gap. This is furthered when it comes to the social, economic, and ethical consequences of AI's rapid adoption which is leading to (I) economic losses, due to declining jobs and a decline in GDP growth (countries which have low value-add exports, or are behind on IT in general battle with this more), coupled with (II) a mistrust, misunderstanding, and no "check and balance" (in particular for AI applications in Media, Defence, et al.). Transparency, accountability and human oversight needs to be endorsed and structured at the United Nations level. The United Nations (via the ITU) or other bodies must build such a global oversight mechanism, with a defined set of rules and timed commitments (similar to the UNFCCC, and the Paris Agreement's approach) and additionally provide a finance mechanism which can provide capital and technology (open source models, open data, and implementation techniques) to developing countries to achieve such targets. In addition to this as highlighted in emprical studies the quality of AI models is heavility dependent upon: (I) the data-sets, repositories etc. that models are trained on and (II) the humans that provide reinforcement learning to AI models. For developing countries, and for the MENAP region where there is a lack of availablity of local data, no training of models based on local data and (III) no humans to re-inforce models this creates a strong bias, that such models will aways favour or provide accuracy purely on use-cases, or data sets from developing countries. In order to solve the above the General Assembly Resolution 79/325, must be accompanied by a a globally acceptable Agreement, which encapsulates in a frameowork such as the United Nations Framework in Artificial Intelligence (UNFAI), which follows the principles of accuracy, transparency and equitable economic growth for developed and developing countries.

For the MENAP region and developing regions in particular, AI capacity building and understanding of the technology is the biggest fundamental challenge. For developing countries across the region, a lack of underpinning regulation on data protection is the missing gap. This is furthered when it comes to the social, economic, and ethical consequences of AI's rapid adoption which is leading to (I) economic losses, due to declining jobs and a decline in GDP growth (countries which have low value-add exports, or are behind on IT in general battle with this more), coupled with (II) a mistrust, misunderstanding, and no "check and balance" (in particular for AI applications in Media, Defence, et al.). Transparency, accountability and human oversight needs to be endorsed and structured at the United Nations level. The United Nations (via the ITU) or other bodies must build such a global oversight mechanism, with a defined set of rules and timed commitments (similar to the UNFCCC, and the Paris Agreement's approach) and additionally provide a finance mechanism which can provide capital and technology (open source models, open data, and implementation techniques) to developing countries to achieve such targets. In addition to this as highlighted in emprical studies the quality of AI models is heavility dependent upon: (I) the data-sets, repositories etc. that models are trained on and (II) the humans that provide reinforcement learning to AI models. For developing countries, and for the MENAP region where there is a lack of availablity of local data, no training of models based on local data and (III) no humans to re-inforce models this creates a strong bias, that such models will aways favour or provide accuracy purely on use-cases, or data sets from developing countries. In order to solve the above the General Assembly Resolution 79/325, must be accompanied by a a globally acceptable Agreement, which encapsulates in a frameowork such as the United Nations Framework in Artificial Intelligence (UNFAI), which follows the principles of accuracy, transparency and equitable economic growth for developed and developing countries.

Pakistan & Egypt (I) No data protection law - no firm regulation for safe use of data. (II) No development financing - no local compute infrastructure, energy, industry or talent to build. (III) IT Exports - mainly low value add, decreasing. (IV) Loss of Jobs - in particular due to AI, robotics and industrial use cases as Pakistan has blue collar labour which is impacted hence leading to a decreased in remittances and a worse current account definicit (implying more funding and dependency on IMF and WB bodies for loans). Saudi Arabia (I) Lack of local AI model development, data for training et al. - mistrust of AI models. (II) Lack of AI talent in policy-making and regulators - KSA does have SADIA, yet capacity, understanding and ability to reinforce and localise the technology is a gap.