If your organisation uses AI to answer questions about Suriname for Dutch-speaking users, check whether it can work accurately in Dutch. UNESCO's national readiness assessment offers a reason to gather that evidence now.
Disclosure. The Caribbean AI Association was founded by Adrian Dunkley, and this site is supported by StarApple AI, which describes itself as the Caribbean's first AI company. Neither is a party to the Suriname assessment, and no fact about the Suriname assessment is drawn from either.
The 4th UNESCO Global Forum on the Ethics of AI ran from 14 to 17 September 2026 in Saudi Arabia, co-hosted by UNESCO with the Saudi Data and AI Authority and the International Centre for AI Research and Ethics. UNESCO Kingston's note, published on 23 September and updated on 24 September, says the Bahamas, Saint Vincent and the Grenadines, Suriname, and Trinidad and Tobago contributed to discussions on AI readiness, public-sector skills, social inequality and environmental sustainability. Members of the AI Ethics Action Cohort from Curaçao, Suriname, and Trinidad and Tobago received expert training on UNESCO's AI ethics tools and explored how to apply them in their national contexts. Devdiscourse re-reported the note on 24 September.
The forum report connects Suriname's participation to its recent national assessment launch. It describes Baasaron's view that RAM can help the country assess its own position and learn from other countries.
RAM stands for Readiness Assessment Methodology. It is one of the tools UNESCO built around its Recommendation on the Ethics of Artificial Intelligence, which member states adopted in 2021. UNESCO's AI page says that as of October 2025 over 75 countries from five continents had initiated or completed RAM assessments. Before Suriname, UNESCO's Kingston office lists Antigua and Barbuda, Curaçao, Jamaica, and Trinidad and Tobago as Caribbean countries that had taken a RAM path.
Suriname's assessment will look at where the country stands on governance, regulation, education and infrastructure, and what still needs to be built. UNESCO says the next steps are surveys, interviews and sector consultations on how AI is already used in Suriname, where the opportunities are, and which governance, skills and safeguards are needed. The work is part of a UNESCO project funded by the Japanese Funds-in-Trust.
Devdiscourse's report states: "not a final policy or a simple technology checklist." The wording is Devdiscourse's. The assessment records the country's position and gaps. Any new law would require a separate government and legislative process.
A readiness assessment needs a baseline. These are the public ones we could check at source.
Meeting size: 106 participants, 67 institutions, 10 sectors. Source: UNESCO, "Suriname Advances Responsible AI Governance with UNESCO's RAM 2.0", 2 September 2026. Our arithmetic: that is 1.6 participants per institution, and 106 people is 0.017% of the population.
Population: 639,850 in 2025. Source: World Bank, indicator SP.POP.TOTL. A first meeting of 106 says nothing about the quality of consultation, only about its scale.
Individuals using the internet: 87.4% of the population in 2023, up from 45.4% in 2016. Source: World Bank, indicator IT.NET.USER.ZS. The rise is close to double, on our arithmetic.
Mobile cellular subscriptions: 142.1 per 100 people in 2024, down from 157.2 in 2023. Source: World Bank, indicator IT.CEL.SETS.P2. Subscriptions are SIM accounts, not users, so a figure above 100 says nothing about how many individuals are online.
Real GDP growth: 1.5% in 2025, with 3.9% projected for 2026. Source: IMF DataMapper, series NGDP_RPCH, retrieved 12 October 2026. GDP per capita was US$7,070 in 2025 (World Bank, NY.GDP.PCAP.CD).
Read together, these describe a small country that is already largely online. Use of AI tools has not been measured in any of these sources, and we found no Surinamese adoption survey for this article. That is the kind of gap the assessment is meant to fill.
For Surinamese organisations the practical meaning is timing. The assessment has opened, the survey and interview stage is ahead, and the evidence an organisation brings to it will count for more than an opinion. For readers elsewhere in the Caribbean, Suriname is a working example of starting with a national audit of governance, education and infrastructure, and Jamaica and Trinidad and Tobago have already taken the same route.
Dutch is Suriname's official language. The Dutch Senate's account of a 2024 ceremony for the Nederlandse Taalunie says the Netherlands, Flanders and Suriname cooperate through that body on the Dutch language, and quotes a Surinamese minister, named there as Ori, in Dutch: "Nederlands, hoewel de officiële taal, concurreert bij ons met 22 lokale talen." In English: Dutch, although the official language, competes in Suriname with 22 local languages. The page does not state the minister's portfolio.
An AI tool needs separate checks for questions about Suriname in English, Dutch, Sranan Tongo and other local languages. A tool can pass the first and fail the second. Nothing in the sources we read tests any of them. The Bahamas' Wayde A. Watson raised the same family of concern at the forum: little Caribbean history, legal knowledge and spoken language appear in AI training material, in Devdiscourse's report of the note. For Suriname, a test conducted only in English would leave the Dutch-language question unanswered.
The answer key should also be Surinamese. The ABS, Suriname's General Bureau of Statistics, publishes at statistics-suriname.org, and its news page records recent facts a model may not hold, such as the official welcome of Dr Barbara Adams as ABS director on Monday 17 August 2026. Its census page states that the National Assembly adopted the law for the Ninth General Population and Housing Census (Negende Algemene Volkstelling) on 4 October 2024.
Adrian Dunkley's library of methods includes Preparation Asymmetry, which describes the structural gap between nations that build AI systems and nations that inherit them. In that framing, Caribbean governments mostly sit in the inheritance position, so rules, training data and model values arrive from elsewhere.
We are applying the method to Suriname here; UNESCO and the government have not made this connection. An organisation choosing an imported tool can first define what it needs to know about local language and law, then test the product against those needs. A national readiness assessment can help identify the questions.
The method also tells you what to check. If the inherited system was trained elsewhere, you cannot assume it knows your country, and in Suriname that includes its language. Nothing in the sources says how well any AI tool answers questions about Suriname in Dutch, so that is a testable gap, and the steps below are built around it.
UNESCO's forum note records two other Caribbean interventions worth keeping next to Suriname's. Wayde A. Watson, the Bahamas' Minister of State for Innovation and National Development, raised concern that little Caribbean history, legal knowledge and spoken language appear in AI training material. Lavern V. King, Saint Vincent and the Grenadines' Minister of State for Education, Innovation, Digital Transformation and Information, said AI should support teachers and strengthen students' critical thinking.
These remarks raise questions for an assessment: which local material do models know, and how are teachers and students using them? UNESCO also says the Caribbean AI Policy Roadmap, endorsed by CARICOM's Council for Trade and Economic Development, was highlighted at the forum. For the regional record on that endorsement, see our earlier article, CARICOM Just Endorsed a Regional AI Roadmap, and for the coordination argument, CARICOM AI Governance: One Voice or Fifteen?
These are suggestions from us. UNESCO sets no requirements here.
The consultation method and the tests an organisation chooses will affect whose experience is recorded.
1. The offline and under-connected are missed. Harm: a survey and consultation process reaches the connected first. Who bears it: residents outside the 87.4% using the internet, which leaves 12.6% by our arithmetic, roughly 80,000 people if that share is applied to the 2025 population (mixing 2023 and 2025 data). Mobile subscriptions also fell from 157.2 to 142.1 per 100 people between 2023 and 2024 (World Bank), and the source gives no reason. Evidence: the World Bank figures and UNESCO's statement that surveys, interviews and sector consultations follow. Inference: that the offline are the group most likely to be missed; we have seen neither the attendee list nor the method. Guardrail: ask the ministry for the attendee sector breakdown and how residents without reliable connections will be reached.
2. A Dutch-only process leaves out the other languages. Harm: consultations and AI tests run in Dutch alone say little about speakers of Sranan Tongo or other local languages. Who bears it: those speakers, who may also be the people an AI tool serves worst. Evidence: the Dutch Senate's report of the minister's remark that Dutch competes with 22 local languages. Uncertainty: UNESCO's pages do not say which languages will be used. We have no evidence that the assessment will be Dutch-only. Guardrail: ask which languages the surveys will be offered in, and record which language each of your own test answers used.
3. A readiness picture is read as a rule. Harm: organisations assume safeguards exist because an assessment is under way. Who bears it: people affected by AI use in services and workplaces where no rule has changed. Evidence: Devdiscourse's report states the assessment is "not a final policy or a simple technology checklist"; UNESCO's account covers an opening meeting and no findings have been published. Inference: that some organisations will lean on the process as cover. Guardrail: keep your own written AI-use rule, and ask for the dated outputs of the assessment as they appear.
We could not confirm a timeline for the assessment, who sits on its steering group, or whether the final report will be public. We found no new Suriname-specific development dated after 12 September beyond the forum note. UNESCO's Suriname article shows a last-updated date of 7 October 2026, and we could not see what changed. We also could not find a Suriname-specific measure of AI use, which is why the numbers above stop at connectivity. This article leans on UNESCO's own account of its own programme, and an outside check of what the assessment finds will have to wait for its report.
Keep a record of the questions your tool got wrong, the sources you checked and any differences between its Dutch and English answers. Ask the ministry how to submit those findings. That gives the consultation team something specific to investigate without treating a small test as a national result.
UNESCO says it will cover where Suriname stands on governance, regulation, education and infrastructure, and what remains to be developed. Surveys, interviews and sector consultations follow, on current AI use, opportunities by sector, and the governance, skills and safeguards needed. The project is funded through the Japanese Funds-in-Trust.
UNESCO's Kingston office lists Antigua and Barbuda, Curaçao, Jamaica, and Trinidad and Tobago as having taken a comprehensive RAM path before Suriname. Globally, UNESCO's AI page says over 75 countries across five continents had initiated or completed assessments as of October 2025, and UNESCO's article describes Suriname as the first in the English- and Dutch-speaking Caribbean to undertake the updated RAM 2.0.
The fourth forum ran from 14 to 17 September 2026 and was co-hosted by UNESCO, the Saudi Data and AI Authority and the International Centre for AI Research and Ethics. UNESCO Kingston's note says the Bahamas, Saint Vincent and the Grenadines, Suriname, and Trinidad and Tobago contributed, and that cohort members from Curaçao, Suriname and Trinidad and Tobago were trained on UNESCO's AI ethics tools.
Not directly. Devdiscourse's report states the assessment is "not a final policy or a simple technology checklist." The wording is Devdiscourse's. Our reading is that it maps the current position and gaps, and that any legislation or national strategy would be a separate decision for government. The 2021 UNESCO Recommendation that underpins RAM was adopted by member states as a framework, and the assessment is one of the tools built around it.
UNESCO says consultations are coming, but the pages we read name no public sign-up. The opening meeting had 106 participants from 67 institutions across 10 sectors, so the start was weighted towards institutions. Contact the economic affairs ministry, ask which sector consultation fits you and for the sector breakdown of the opening meeting, and bring evidence such as results from the Dutch-language test above.
An online survey can miss people without reliable connections. World Bank data puts internet use at 87.4% in 2023, leaving 12.6% outside that measure. Internet use alone does not establish who will see or complete a consultation survey. Mobile subscriptions fell from 157.2 to 142.1 per 100 people between 2023 and 2024, a drop worth watching, though the source does not explain it.
Dutch is Suriname's official language, and the Netherlands, Flanders and Suriname cooperate on it through the Nederlandse Taalunie, according to the Dutch Senate's account of a 2024 ceremony. A Surinamese minister is quoted there saying Dutch competes with 22 local languages. The ABS pages we read use Dutch. A tool serving Dutch-speaking users needs to be checked on Dutch-language questions, alongside tests in the other languages it serves.