LLMs and Agentic AI in Insurance Decision-Making: Opportunities and Challenges For Africa

Graham Hill; James Gachomo Wanjiku; JingYuan Gong; Moseli Mots'oehli; Thulani Babeli

REVIEW 3 major objections 4 minor 100 references

LLMs and Agentic AI in Insurance Decision-Making: Opportunities and Challenges For Africa

T0 review · 3 major / 4 minor · reviewed 2026-08-05 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read LLMs and agentic AI can change African insurance decision-making, the paper argues.

desk verdict A useful Africa-focused perspective on LLMs in insurance, but the submitted full text is corrupt, so only the abstract is evaluable and the body's evidence cannot be checked. read the letter →

arxiv 2508.15110 v1 pith:4MS72DGJ submitted 2025-08-20 cs.CE cs.CLcs.ETstat.AP

Graham Hill , JingYuan Gong , Thulani Babeli , Moseli Mots'oehli , James Gachomo Wanjiku This is my paper

classification cs.CEcs.CLcs.ETstat.AP

keywords largelanguagemodelsagenticAIinsuranceAfricafinancialinclusiongovernanceopen-sourcemarketgaps

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper argues that large language models and agentic AI systems have reached a point where they can meaningfully improve insurance decision-making, and that Africa is positioned to benefit if the technology is built locally. It maps the specific gaps in the African insurance market, points to existing local efforts and partnership opportunities, and calls on actuaries, insurers, regulators, and technology leaders to collaborate on AI strategies that are by and for Africans. The central bet is that rapid performance gains, open-source access, and falling deployment costs create a window for Africa to develop its own AI-driven insurance tools rather than import them. A sympathetic reader would care because the paper frames AI adoption as a development strategy with concrete consequences for coverage, cost, and fairness.

What carries the argument

The enabling mechanism is the convergence of three trends the paper identifies: rapid improvement in LLM and agentic AI performance, growing open-source access, and falling deployment costs. The operative vehicle is the African insurance market itself, treated as a context of underinsurance, local innovation, and partnership opportunities where these capabilities can be directed. Agentic AI—AI that plans and executes multi-step decisions—is the specific capability the paper bets on for insurance decision-making.

What would settle it

A controlled pilot in one African insurance market comparing AI-assisted underwriting and claims handling against the existing human process: if the AI system does not reduce cost per policy or claim, does not cut decision times, or produces error rates no better than the status quo, the paper's central opportunity claim is weakened.

Watch

Extended reading notes

Core claim

The paper's central claim is that large language models and agentic AI, defined as AI systems that can plan and carry out multi-step tasks, are now capable of supporting insurance decisions such as underwriting, claims, and customer service, and that the African insurance market has distinctive gaps these tools are well suited to fill. It asserts that rapid performance improvements, increased open-source access, and decreasing deployment costs make this feasible for African institutions, and that the right path is collaborative, inclusive, sustainable, and equitable AI development driven by local actors. The intended outcome is not just adoption but the creation of insurance AI solutions by

Load-bearing premise

The claim rests on the assumption that LLM and agentic AI systems are currently reliable, affordable, and adaptable enough for real insurance decisions in African contexts, and that the market gaps the paper describes are accurately characterized.

Editorial extensions

If this is right

If the paper is right, African insurers can deploy AI-assisted underwriting and claims tools at costs low enough to reach currently uninsured populations.
Open-source models would let African institutions retain control over data and system design instead of depending on foreign proprietary platforms.
Regulators and actuaries would need to co-design standards for AI fairness, transparency, and local-language performance rather than importing external rules.
Local pilots and partnerships would become the primary unit of progress, with success measured by coverage gains and equitable access.
The actuary's role would shift from building models to supervising and questioning AI outputs in decision-critical settings.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

A natural test the paper does not run: side-by-side pilots in two African markets comparing AI-assisted underwriting and claims against existing human processes on cost, speed, error rate, and customer trust; the paper's thesis predicts measurable gains within a year or two.
The argument implies a leapfrog logic similar to mobile money: Africa may skip legacy insurance IT infrastructure and move straight to agentic, mobile-first distribution, an implication the authors gesture at but do not develop.
A hidden risk the paper does not quantify: if local training data and African-language benchmarks are scarce, model errors and bias could amplify rather than reduce exclusion, so building evaluation data is the first concrete step the framework implies.
The same 'by and for locals' design principle could be tested in other underserved regions to determine whether local ownership is the load-bearing ingredient or merely a beneficial add-on.

Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

Short version: this is a coherent perspective that adds an Africa-specific lens to the LLM-in-insurance conversation, but the submitted full text is unreadable — it even carries a header from a cond-mat paper — so I could only evaluate the abstract. On that basis it is a reasonable call to action, not a research result.

What it does well: the paper frames opportunities and challenges without overselling. It names critical gaps in African insurance, points to local efforts and players, and asks actuaries, insurers, and regulators to collaborate. That framing is useful, especially for a policy-oriented audience, and it's honest about complexity — no benchmark claims, no fitted models.

Where it is soft: the key premises — that LLM/agentic systems are reliable enough for insurance decisions, that deployment costs are falling fast enough for African markets, and that the described gaps are accurate — are asserted in the abstract without supporting evidence. For a perspective that's tolerable, but it means the paper's practical weight depends entirely on the body. And I can't check the body: the supplied text is corrupted, and the recognizable fragments point to a different arXiv paper. If that's a pipeline artifact, fine, but as it stands I have no way to verify the local inventory, any cost figures, or error rates. Also, novelty is modest: the contribution is the regional lens and a list of actors, not a new framework or dataset.

Who it's for: actuaries, insurers, regulators, and researchers working on African fintech/insurance. A scientific reader looking for new results won't find them. I would not cite it in my own work, but I'd send a clean version to a peer-reviewed applied venue — the topic is underserved and a good referee can push the authors to substantiate the feasibility claims.

Recommendation: ask for a clean PDF, then send it to peer review at a policy- or applications-oriented journal. Desk rejecting the idea would be a mistake; desk rejecting this corrupted artifact would be understandable.

Referee Report

3 major / 4 minor

Summary. This paper is a perspective/position piece arguing that LLMs and agentic AI have transformative potential for insurance decision-making, with a particular focus on the African market. The abstract asserts rapid performance improvements, decreasing deployment costs, open-source access, and identifies critical gaps, local efforts, and partnership opportunities, culminating in a call for collaborative, African-led AI strategies. The supplied full text, however, is largely unreadable mojibake and appears to contain a different paper (arXiv:2508.15108v2, cond-mat.mtrl-sci) with materials-science equations. As a result, the body of the claimed insurance paper cannot be assessed, and none of the abstract's empirical premises can be verified from the submitted artifact.

Significance. If the intended full text were available, the paper could serve as a useful agenda-setting piece for AI adoption in African insurance. The stated emphasis on local collaboration and 'by and for Africans' is a constructive framing. The paper does not offer machine-checked proofs, reproducible code, parameter-free derivations, or falsifiable predictions; its value would depend on the quality and accuracy of its contextual analysis, market description, and cited evidence. Given that the abstract itself contains no data, benchmarks, or sources, the significance of the contribution cannot currently be evaluated. The submission format issue is the dominant obstacle.

major comments (3)

[Full text (all pages after the abstract)] The body supplied for review is not the manuscript described by the title and abstract. The visible text is preceded by corrupted Cyrillic-like mojibake, and later pages carry the header 'arXiv:2508.15108v2 [cond-mat.mtrl-sci]' with equations from materials science. No readable section on African insurance gaps, local actors, or partnership opportunities is present. This is load-bearing: every factual claim in the abstract -- market gaps, players, partnership opportunities, cost/reliability trends -- must be checked against the body, and none of that body is assessable. The authors should be asked to resubmit the correct, legible full text.
[Abstract and central claim] The central call to action rests on several empirical premises that are asserted but not evidenced: that current LLM/agentic AI systems are reliable enough for insurance decision-making, that deployment costs are decreasing in African contexts, and that the described gaps and opportunities are accurately characterized. No error rates, cost figures, pilot examples, or references are supplied in the abstract. If the corrected body does not substantiate these premises with referenced evidence -- e.g., insurance-specific evaluations, cost data, or documented African deployments -- the recommendation remains an assertion rather than a supported analysis.
[Scope and methodology] The paper promises to 'identify critical gaps in the African insurance market' and 'highlight key local efforts, players, and partnership opportunities,' but no method is stated for selecting these gaps, players, or efforts. For a narrative or position piece, a brief statement of search criteria, data sources, or selection rationale is needed to allow the reader to judge whether the account is representative or selective. This is particularly important because the practical value of the paper depends on the accuracy of its African market characterization.

minor comments (4)

[Full text metadata] The submitted PDF includes the arXiv identifier and header of another paper (arXiv:2508.15108v2, cond-mat.mtrl-sci). This should be removed and replaced with the actual manuscript metadata. The date header '1 Apr 2026' is also inconsistent with the submitted arXiv ID and should be checked.
[References] No readable reference list is present in the supplied artifact. The corrected version should include full citations for all claims about AI capability, cost trends, open-source availability, and African insurance market conditions.
[Title and framing] The title promises 'Opportunities and Challenges,' but the abstract emphasizes opportunities and pathways and does not name a single concrete challenge (e.g., hallucination risk, bias, data scarcity, regulatory capacity). The corrected body should give explicit, balanced treatment of both sides.
[Claim about collaboration] The closing call for inclusive and equitable AI strategies is reasonable, but it should be connected to concrete stakeholder responsibilities or policy levers. As written, it is a vision statement rather than an operational recommendation.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the paper is a narrative/perspective piece with no derivation chain, and the supplied full text is corrupted/unreadable, so no circular reduction can be exhibited.

full rationale

The abstract makes an argumentative claim about the potential of LLMs and agentic AI in African insurance and calls for collaboration; it does not present a derivation, fit, or prediction whose output is defined by its inputs. No equations, fitted parameters, or load-bearing self-citations are present in the readable portion. The supplied full text is largely unreadable mojibake and appears to contain material from a different arXiv submission (arXiv:2508.15108v2, cond-mat.mtrl-sci), so no specific reduction such as Eq. X = Eq. Y by construction can be quoted. Under the hard rule requiring quotation and exhibiting a specific reduction, no circularity can be identified. The lack of verifiable evidence about the body's empirical premises is an evidence/correctness concern, not a circularity finding.

Assumptions & free parameters 0 free parameters · 3 assumptions · 0 invented entities

Since the paper is a perspective piece, the assumptions are contextual judgments about the AI landscape and the African market, rather than mathematical axioms. No free parameters or invented entities are introduced.

assumptions (3)

domain assumption LLM and agentic AI capabilities will continue to improve and deployment costs will continue to decrease.
The paper's premise of 'rapid performance improvements, increased open-source access, decreasing deployment costs' is stated in the abstract but not demonstrated.
domain assumption The identified gaps in the African insurance market are real and can be addressed by AI-based solutions.
The abstract asserts critical gaps and local opportunities without presenting supporting data in the abstract.
domain assumption Stakeholder collaboration can produce inclusive, sustainable, and equitable AI strategies in Africa.
The call to action presumes that regulatory and institutional collaboration is feasible and sufficient.

how reviews work

0 comments

Cite this review

Pith. "Pith review of LLMs and Agentic AI in Insurance Decision-Making: Opportunities and Challenges For Africa." pith.science (2026). https://pith.science/paper/4MS72DGJ

@misc{pith2026250815110,
  author       = {Pith},
  title        = {Pith review of: LLMs and Agentic AI in Insurance Decision-Making: Opportunities and Challenges For Africa},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/4MS72DGJ}},
  note         = {Machine review of arXiv:2508.15110}
}

read the original abstract

In this work, we highlight the transformative potential of Artificial Intelligence (AI), particularly Large Language Models (LLMs) and agentic AI, in the insurance sector. We consider and emphasize the unique opportunities, challenges, and potential pathways in insurance amid rapid performance improvements, increased open-source access, decreasing deployment costs, and the complexity of LLM or agentic AI frameworks. To bring it closer to home, we identify critical gaps in the African insurance market and highlight key local efforts, players, and partnership opportunities. Finally, we call upon actuaries, insurers, regulators, and tech leaders to a collaborative effort aimed at creating inclusive, sustainable, and equitable AI strategies and solutions: by and for Africans.

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

100 extracted references · 68 canonical work pages

[1]

N. N. Mabaso, ``Evaluation of the implementation of the twin peaks financial sector regulatory model in south africa,'' Ph.D. dissertation, University of Pretoria, 2021. [Online]. Available: https://repository.up.ac.za/bitstreams/106d7444-3186-4884-ae40-53b858d8d611/download

2021
[2]

Seiranian and A

M. Seiranian and A. Suzuki, ``Large language model parameter efficient fine-tuning for mathematical problem solving,'' Ph.D. dissertation, 07 2024

2024
[3]

Adversarial AI in Insurance: Pervasiveness and Resilience

E. Luciano, M. Cattaneo, and R. Kenett, ``Adversarial ai in insurance: Pervasiveness and resilience,'' arXiv preprint arXiv:2301.07520, 2023

work page Pith review arXiv 2023
[4]

63, 2025

Bank for International Settlements , ``Regulating ai in the financial sector: Recent developments and main themes,'' BIS FSI Insights Paper No. 63, 2025. [Online]. Available: https://www.bis.org/fsi/publ/insights63.pdf

2025
[5]

C. E. Shannon, ``Prediction and entropy of printed english,'' The Bell System Technical Journal, vol. 30, no. 1, pp. 50--64, 1951

1951
[6]

Hochreiter and J

S. Hochreiter and J. Schmidhuber, ``Long short-term memory,'' Neural Comput., vol. 9, no. 8, p. 1735–1780, 1997

1997
[7]

Chung, C

J. Chung, C. Gulcehre, K. Cho, and Y. Bengio, ``Empirical evaluation of gated recurrent neural networks on sequence modeling,'' arXiv preprint arXiv:1412.3555, 2014

arXiv 2014
[8]

J. D. Owens, M. Houston, D. Luebke, S. Green, J. E. Stone, and J. C. Phillips, ``Gpu computing,'' Proceedings of the IEEE, vol. 96, no. 5, pp. 879--899, 2008

2008

Show all 100 references

[9]

Mikolov, M

T. Mikolov, M. Karafi \'a t, L. Burget, J. C ernock \'y , and S. Khudanpur, ``Recurrent neural network based language model,'' in INTERSPEECH, 2010, pp. 1045--1048

2010
[10]

Brants and A

T. Brants and A. Franz, ``Web 1t 5-gram version 1,'' Linguistic Data Consortium, LDC2006T13, 2006

2006
[11]

Jozefowicz, O

R. Jozefowicz, O. Vinyals, M. Schuster, N. Shazeer, and Y. Wu, ``Exploring the limits of language modeling,'' arXiv preprint arXiv:1602.02410, 2016. [Online]. Available: https://arxiv.org/abs/1602.02410

2016 arXiv
[12]

Vaswani, N

A. Vaswani, N. Shazeer, N. Parmar, J. Uszkoreit, L. Jones, A. N. Gomez, . Kaiser, and I. Polosukhin, ``Attention is all you need,'' in Advances in Neural Information Processing Systems, vol. 30, 2017, pp. 5998--6008

2017
[13]

Radford, K

A. Radford, K. Narasimhan, T. Salimans, and I. Sutskever, ``Improving language understanding by generative pre-training,'' arXiv preprint arXiv:1801.06146, 2018

2018 arXiv
[14]

Radford, J

A. Radford, J. Wu, R. Child, D. Luan, D. Amodei, and I. Sutskever, ``Language models are unsupervised multitask learners,'' OpenAI blog, 2019, https://cdn.openai.com/better-language-models/language_models_are_unsupervised_multitask_learners.pdf

2019
[15]

T. B. Brown, B. Mann, N. Ryder, M. Subbiah, J. Kaplan, P. Dhariwal, A. Neelakantan, P. Shyam, G. Sastry, A. Askell, S. Agarwal, A. Herbert-Voss, G. Krueger, T. Henighan, R. Child, A. Ramesh, D. M. Ziegler, J. Wu, C. Winter, C. Hesse, M. Chen, E. Sigler, M. Litwin, S. Gray, B. ...

2020
[16]

Hendrycks, C

D. Hendrycks, C. Burns, S. Basart, A. Zou, M. Mazeika, D. Song, and J. Steinhardt, ``Measuring massive multitask language understanding,'' Proceedings of the International Conference on Learning Representations (ICLR), 2021

2021
[17]

E. M. Bender, T. Gebru, A. McMillan-Major, and M. Shmitchell, ``On the dangers of stochastic parrots: Can language models be too big?'' in Proceedings of the 2021 ACM Conference on Fairness, Accountability, and Transparency. 1em plus 0.5em minus 0.4em ACM, 2021, pp. 610--623

2021
[18]

Reynolds and K

L. Reynolds and K. McDonell, ``Prompt programming for large language models: Beyond the few-shot paradigm,'' in Extended Abstracts of the 2021 CHI Conference on Human Factors in Computing Systems, 2021

2021
[19]

Touvron, L

H. Touvron, L. Martin, K. Stone, P. Albert, M. Douze, N. Goyal, T. Raya, C. Raffel, and A. Joulin, ``Llama: Open and efficient foundation language models,'' 2023

2023
[20]

OpenAI , `` GPT-4 technical report,'' https://cdn.openai.com/papers/gpt-4.pdf, 2024

2024
[21]

Anthropic , ``Introducing claude 2: A next‐generation conversational llm,'' 2023, https://www.anthropic.com/product-overview/claude-2

2023
[22]

DeepSeekR Team , ``Deepseekr1: A multimodal retrieval‐augmented foundation model,'' 2023, https://github.com/deepseekr/deepseekr1

2023
[23]

Stiennon, L

N. Stiennon, L. Ouyang, J. Wu, D. M. Ziegler, R. Lowe, C. Voss, A. Radford, D. Amodei, and P. Christiano, ``Learning to summarize with human feedback,'' arXiv preprint arXiv:2009.01325, 2020

2009 arXiv
[24]

Lepikhin, H

T. Lepikhin, H. Lee, Y. Xu, T. Chen, O. Firat, M. Huang, A. Jolicoeur‐Martineau, and N. Shazeer, ``Gshard: Scaling giant models with conditional computation and automatic sharding,'' in International Conference on Learning Representations, 2020. [Online]. Available: https://ar...

2020 arXiv
[25]

J. Wei, X. Wang, D. Schuurmans, M. Bosma, B. Ichter, F. Xia, E. H. Chi, Q. V. Le, and D. Zhou, ``Chain-of-thought prompting elicits reasoning in large language models,'' in Proceedings of the 36th International Conference on Neural Information Processing Systems, 2022

2022
[26]

W. Wang, F. Wei, L. Dong, H. Bao, N. Yang, and M. Zhou, ``Minilm: Deep self-attention distillation for task-agnostic compression of pre-trained transformers,'' Advances in neural information processing systems, vol. 33, pp. 5776--5788, 2020

2020
[27]

Huang, D

S. Huang, D. Siddarth, L. Lovitt, T. I. Liao, E. Durmus, A. Tamkin, and D. Ganguli, ``Collective constitutional ai: Aligning a language model with public input,'' in Proceedings of the 2024 ACM Conference on Fairness, Accountability, and Transparency, 2024, p. 1395–1417

2024
[28]

S. Wu, O. Irsoy, S. Lu, V. Dabravolski, M. Dredze, S. Gehrmann, P. Kambadur, D. Rosenberg, and G. Mann, ``Bloomberggpt: A large language model for finance,'' arXiv preprint arXiv:2303.17564, 2023

2023 arXiv
[29]

J. Ding, K. Feng, B. Lin, J. Cai, Q. Wang, Y. Xie, X. Zhang, Z. Wei, and W. Chen, ``Insqabench: Benchmarking chinese insurance domain question answering with large language models,'' arXiv preprint arXiv:2501.10943, 2025

2025 arXiv
[30]

R. Luo, L. Sun, Y. Xia, T. Qin, S. Zhang, H. Poon, and T.-Y. Liu, ``Biogpt: Generative pre-trained transformer for biomedical text generation and mining,'' in arXiv preprint arXiv:2210.10341, 2022

2022 arXiv
[31]

D. Shu, H. Zhao, X. Liu, D. Demeter, M. Du, and Y. Zhang, ``Lawllm: Law large language model for the us legal system,'' arXiv preprint arXiv:2407.21065, 2024

2024 arXiv
[32]

Martin, B

L. Martin, B. Muller, V. Su \'e ry, \'E . d. l. C. Marchand, C. Soussen, and C. Darve, ``Camembert: a tasty french language model,'' in Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics. 1em plus 0.5em minus 0.4em Association for Computati...

2020
[33]

Cañete and J

S. Cañete and J. Tiedemann, ``Spanish‐bert: A large‐scale language model for spanish,'' 2021, https://github.com/dccuchile/beto

2021
[34]

Y. Cui, W. Che, T. Liu, and G. Hu, ``Chinese‐bert‐wwm: Whole word masking for chinese bert,'' 2019, https://github.com/ymcui/Chinese-BERT-wwm

2019
[35]

Kossaï, M

B. Kossaï, M. Zeggour, J. Mpiana, and V. Umapathi, ``Afriberta: A bert‐based language model for african languages,'' in Proceedings of the Second Workshop on Language Technologies for African Languages, 2022, pp. 1--8

2022
[36]

S. Yao, J. Zhao, D. Yu, N. Du, I. Shafran, K. Narasimhan, and Y. Cao, `` ReAct : Synergizing reasoning and acting in language models,'' in International Conference on Learning Representations (ICLR), 2023

2023
[37]

Liu and B

J. Liu and B. Mozafari, ``Query rewriting via large language models,'' arXiv preprint arXiv:2403.09060, 2024

2024
[38]

Chase, ``Langchain: A framework for developing applications with large language models,'' https://github.com/hwchase17/langchain, 2023

H. Chase, ``Langchain: A framework for developing applications with large language models,'' https://github.com/hwchase17/langchain, 2023

2023
[39]

Kwok et al., ``Llamaindex (gpt index): A data framework for your llm application,'' https://github.com/run-llama/llama_index, 2023

J. Kwok et al., ``Llamaindex (gpt index): A data framework for your llm application,'' https://github.com/run-llama/llama_index, 2023

2023
[40]

Andersen, M

C. Andersen, M. Shaddad, D. Lemaire et al., ``Microsoft semantic kernel: An orchestration sdk for large language models,'' https://github.com/microsoft/semantic-kernel, 2023

2023
[41]

Salman et al., ``Auto-gpt: Autonomous ai agents powered by gpt-4 and beyond,'' https://github.com/Significant-Gravitas/Auto-GPT, 2023

S. Salman et al., ``Auto-gpt: Autonomous ai agents powered by gpt-4 and beyond,'' https://github.com/Significant-Gravitas/Auto-GPT, 2023

2023
[42]

Huryn, ``The ultimate guide to ai agents for pms,'' The Product Compass Premium newsletter, jun 2025, accessed: 2025-07-03

P. Huryn, ``The ultimate guide to ai agents for pms,'' The Product Compass Premium newsletter, jun 2025, accessed: 2025-07-03. [Online]. Available: https://www.productcompass.pm/p/ai-agents

2025
[43]

McCurry, ``Japanese company replaces office workers with artificial intelligence,'' The Guardian, accessed: 2025-07-03

J. McCurry, ``Japanese company replaces office workers with artificial intelligence,'' The Guardian, accessed: 2025-07-03. [Online]. Available: https://www.theguardian.com/technology/2017/jan/05/japanese-company-replaces-office-workers-artificial-intelligence-ai-fukoku-mutual-...

2025
[44]

Orhan, ``Building high performance document ocr systems,'' Tractable AI Blog, Jun

M. Orhan, ``Building high performance document ocr systems,'' Tractable AI Blog, Jun. 2024, accessed: 2025-07-04. [Online]. Available: https://tractable.ai/building-high-performance-document-ocr-systems/

2024
[45]

[Online]

Tractable , ``Geico partners with tractable to accelerate accident recovery with ai,'' Press release, 2021, accessed: 2025-07-04. [Online]. Available: https://www.prnewswire.com/news-releases/geico-partners-with-tractable-to-accelerate-accident-recovery-with-ai-301299295.html

2021
[46]

Clere, ``Tractable snaps up \ 65m led by softbank for car and property damage appraisals using ai,'' TechCrunch, accessed: 2025-07-04

A. Clere, ``Tractable snaps up \ 65m led by softbank for car and property damage appraisals using ai,'' TechCrunch, accessed: 2025-07-04. [Online]. Available: https://techcrunch.com/2023/07/18/tractable-snaps-up-65m-led-by-softbank-for-car-and-property-damage-appraisals-using-ai

2025
[47]

A. C. Kumar, ``Llm-powered chatbots for customer support in the insurance industry,'' Aidetic Blog, accessed: 2025-07-04. [Online]. Available: https://blog.aidetic.in/llm-powered-chatbots-for-customer-support-in-the-insurance-industry-7c80b0aa3163

2025
[48]

[Online]

Shift Technology , ``About shift technology,'' Company website, 2025, accessed: 2025-07-04. [Online]. Available: https://www.shift-technology.com/about

2025
[49]

[Online]

Shift Technology Claims , ``Ai‐powered claims fraud detection for p&c insurers,'' Company website, 2025, accessed: 2025-07-04. [Online]. Available: https://www.shift-technology.com/resources/research/generative-ai-in-insurance-use-cases-examples-and-real-results

2025
[50]

[Online]

PricewaterhouseCoopers , ``Ai integration in legacy insurance systems: Challenges and solutions,'' PwC Global Report, 2024. [Online]. Available: https://www.pwc.com/ai-insurance-legacy-systems

2024
[51]

[Online]

Deloitte , ``Generative ai in insurance: Navigating costs and maintenance,'' Deloitte Insights, 2025. [Online]. Available: https://www.deloitte.com/insights/insurance-generative-ai

2025
[52]

Pretorius and A

L. Pretorius and A. Bosch, ``Deepmorph: A deep learning approach to morphological analysis for south african bantu languages,'' Computational Linguistics in Africa, vol. 12, no. 1, pp. 45--60, 2020

2020
[53]

Marivate, M

V. Marivate, M. Mots’öehli, V. Wagner, R. Lastrucci, and I. Dzingirai, ``Puoberta: Training and evaluation of a curated language model for setswana,'' in Communications in Computer and Information Science, Artificial Intelligence Research. 1em plus 0.5em minus 0.4em Springer ...

2023 arXiv
[54]

Oseni, N

O. Oseni, N. Adegboye, and F. Lumor, ``Building word embeddings for yoruba using deep learning,'' in Proceedings of the 12th International Conference on Language Resources and Evaluation (LREC), 2020, pp. 3503--3510

2020
[55]

C. Eze, I. Ekwueme, and C. Obi, ``Igbobert: Building a pretrained language model for igbo,'' in Proceedings of the 5th Workshop on African Language Technology, 2022, pp. 45--54

2022
[56]

Abdullahi, I

Z. Abdullahi, I. Musa, and F. Bello, ``Hausabert: A transformer-based pretrained model for hausa,'' in Proceedings of the Workshop on African Language Technology, 2021, pp. 23--32

2021
[57]

K. Uwa, O. Nwachukwu, and P. Okoro, ``Fine-tuning pretrained bert models for named-entity recognition and sentiment analysis in nigerian languages,'' in Proceedings of the 4th Conference on African Language NLP, 2023, pp. 112--121

2023
[58]

Ngugi, S

J. Ngugi, S. Mburu, and E. Mutahi, ``A bidirectional lstm approach to named-entity recognition in swahili,'' in Proceedings of the 10th African Conference on Machine Learning and Data Mining, 2019, pp. 120--131

2019
[59]

D. I. Adelani, D. Babatunde, B. T. Gebre, M. Foufi, T. J. Humphreys, F. Nwachukwu, and D. Msihid, ``Masakhane: Machine translation for african languages,'' in Proceedings of the Third Conference on Machine Translation (WMT), 2021, pp. 497--503

2021
[60]

J. Ojo, K. Ogueji, P. Stenetorp, and D. I. Adelani, ``How good are large language models on african languages?'' arXiv preprint arXiv:2311.07978, 2023

2023 arXiv
[61]

DeWitt Prat, O

L. DeWitt Prat, O. Ndlovu Lucas, C. Golias, and M. Lewis, ``Decolonizing llms: An ethnographic framework for ai in african contexts,'' in Ethnographic Praxis in Industry Conference Proceedings, no. 1, 2024, pp. 46--85

2024
[62]

A. L. Tonja, B. F. Dossou, J. Ojo, J. Rajab, F. Thior, E. P. Wairagala, A. Aremu, P. Moiloa, J. Abbott, V. Marivate et al., ``Inkubalm: A small language model for low-resource african languages,'' arXiv preprint arXiv:2408.17024, 2024

2024 arXiv
[63]

J. O. Alabi, M. A. Hedderich, D. I. Adelani, and D. Klakow, ``Charting the landscape of african nlp: Mapping progress and shaping the road ahead,'' arXiv preprint arXiv:2505.21315, 2025

2025
[64]

(2024) The future of car insurance in south africa: How ai is revolutionising the industry

Pineapple. (2024) The future of car insurance in south africa: How ai is revolutionising the industry. Pineapple Insurance. Blog post. [Online]. Available: https://pineapple.co.za/post/the-future-of-car-insurance-in-south-africa-how-ai-is-revolutionising-the-industry/

2024
[65]

Khanna, ``Kenyan insurtech startup pula raises \ 6m series a,'' TechCrunch, accessed: 2025-07-04

A. Khanna, ``Kenyan insurtech startup pula raises \ 6m series a,'' TechCrunch, accessed: 2025-07-04. [Online]. Available: https://techcrunch.com/2021/01/25/kenyan-insurtech-startup-pula-raises-6m-series-a-led-by-tlcom-capital/

2025
[66]

[Online]

Turaco Insurance , ``About us,'' Company website, 2025, accessed: 2025-07-04. [Online]. Available: https://www.turaco.insure/

2025
[67]

Company, ``Turaco is making insurance more affordable to millions in africa,'' Fast Company, accessed: 2025-07-04

F. Company, ``Turaco is making insurance more affordable to millions in africa,'' Fast Company, accessed: 2025-07-04. [Online]. Available: https://www.fastcompany.com/91037559/turaco-most-innovative-companies-2024

2025
[68]

[Online]

Naked Insurance3 , ``Ai-driven claims automation at naked,'' Naked Insurance blog, 2023, accessed: 2025-07-04. [Online]. Available: https://naked.insure/blog/ai-driven-claims-automation

2023
[69]

Tsanni, ``This company is building ai for african languages,'' MIT Technology Review, 2023, accessed: 2025-07-04

A. Tsanni, ``This company is building ai for african languages,'' MIT Technology Review, 2023, accessed: 2025-07-04. [Online]. Available: https://www.technologyreview.com/2023/11/17/1083637/lelapa-ai-african-languages-vulavula/

2023
[70]

Richman, ``Ai in actuarial science,'' insureAI / University of the Witwatersrand, SSRN Electronic Journal SSRN 3218082, July 2018

R. Richman, ``Ai in actuarial science,'' insureAI / University of the Witwatersrand, SSRN Electronic Journal SSRN 3218082, July 2018. [Online]. Available: https://ssrn.com/abstract=3218082

2018
[71]

Balona, ``Operationalizing llms: A guide for actuaries,'' Actuarial Innovation and Technology Strategic Research Program Technical Report, 2025

C. Balona, ``Operationalizing llms: A guide for actuaries,'' Actuarial Innovation and Technology Strategic Research Program Technical Report, 2025. [Online]. Available: https://www.soa.org/48ffb2/globalassets/assets/files/resources/research-report/2025/oper-genai-act-report.pdf

2025
[72]

------, ``Actuarygpt: Applications of large language models to insurance and actuarial work,'' British Actuarial Journal, vol. 29, p. e15, 2024

2024
[73]

Troxler and J

A. Troxler and J. Schelldorfer, ``Actuarial applications of natural language processing using transformers: Case studies for using text features in an actuarial context,'' British Actuarial Journal, vol. 29, p. e4, 2024

2024
[74]

Dimri, S

A. Dimri, S. Yerramilli, P. Lee, S. Afra, and A. Jakubowski, ``Enhancing claims handling processes with insurance based language models,'' in 2019 18th IEEE international conference on machine learning and applications (ICMLA). 1em plus 0.5em minus 0.4em IEEE, 2019, pp. 1750--1755

2019
[75]

Jani and S

P. Jani and S. Anand, ``Compliance-aware ai adjudication using llms in claims engines (delta lake+ langchain),'' International Journal of Artificial Intelligence, Data Science, and Machine Learning, vol. 5, no. 2, pp. 37--46, 2024

2024
[76]

Krasheninnikova, J

E. Krasheninnikova, J. Garc \' a, R. Maestre, and F. Fern \'a ndez, ``Reinforcement learning for pricing strategy optimization in the insurance industry,'' Engineering applications of artificial intelligence, vol. 80, pp. 8--19, 2019

2019
[77]

Balasubramanian, A

R. Balasubramanian, A. Libarikian, and D. McElhaney, ``Insurance 2030 - the impact of ai on the future of insurance,'' McKinsey & Company, pp. 1--10, 2018
[78]

Chung, C

P. Chung, C. T. Fong, A. M. Walters, N. Aghaeepour, M. Yetisgen, and V. N. O’Reilly-Shah, ``Large language model capabilities in perioperative risk prediction and prognostication,'' JAMA surgery, vol. 159, no. 8, pp. 928--937, 2024

2024
[79]

D. J. Fagnant and K. Kockelman, ``Preparing a nation for autonomous vehicles: opportunities, barriers and policy recommendations,'' Transportation Research Part A: Policy and Practice, vol. 77, pp. 167--181, 2015. [Online]. Available: https://www.sciencedirect.com/science/arti...

2015
[80]

Mots'oehli, F

M. Mots'oehli, F. Chen, H. W. Chan, I. Tlali, T. Babeli, K. Baek, and H. Chen, ``Simulating refractive distortions and weather-induced artifacts for resource-constrained autonomous perception,'' arXiv preprint arXiv:2507.05536, 2025, accepted to the ICCV 2025 Workshop on Compu...

2025 arXiv
[81]

Jiang, Y

F. Jiang, Y. Jiang, H. Zhi, Y. Dong, H. Li, S. Ma, Y. Wang, Q. Dong, H. Shen, and Y. Wang, ``Artificial intelligence in healthcare: Past, present and future,'' Stroke and Vascular Neurology, vol. 2, no. 4, pp. 230--243, 2017

2017
[82]

Topol, Deep Medicine: How Artificial Intelligence Can Make Healthcare Human Again

E. Topol, Deep Medicine: How Artificial Intelligence Can Make Healthcare Human Again. 1em plus 0.5em minus 0.4em New York: Basic Books, 2019

2019
[83]

Reddy, J

S. Reddy, J. Fox, and M. P. Purohit, ``Artificial intelligence-enabled healthcare delivery,'' Journal of the Royal Society of Medicine, vol. 112, no. 1, pp. 22--28, 2019

2019
[84]

Bommasani, D

R. Bommasani, D. A. Hudson, E. Adeli, R. Altman, S. Arora, S. von Arx, M. Bernstein, J. Bohg, A. Bosselut, E. Brunskill et al., ``On the opportunities and risks of foundation models,'' arXiv preprint arXiv:2108.07258, 2021

2021 arXiv
[85]

Information Regulator (South Africa) , ``Protection of personal information act (popi act),'' https://popia.co.za/, n.d
[86]

White & Case LLP , ``Ai watch: Global regulatory tracker - south africa,'' https://www.whitecase.com/insight-our-thinking/ai-watch-global-regulatory-tracker-south-africa, 2024

2024
[87]

Statista , ``Gdp of african countries by country,'' https://www.statista.com/statistics/1120999/gdp-of-african-countries-by-country/, 2024

2024
[88]

Ministry of Information, Communications and the Digital Economy (Kenya) , ``Kenya artificial intelligence strategy 2025–2030,'' https://ict.go.ke/sites/default/files/2025-03/Kenya

2025
[89]

Pinsent Masons , ``South africa takes first steps towards ai regulation,'' https://www.pinsentmasons.com/out-law/news/south-africa-takes-first-steps-towards-ai-regulation, 2024

2024
[90]

Jaldi, ``Artificial intelligence revolution in africa: Economic opportunities and legal challenges,'' Policy Cent

A. Jaldi, ``Artificial intelligence revolution in africa: Economic opportunities and legal challenges,'' Policy Cent. New South, pp. 2023--07, 2023

2023
[91]

M. Raza, Z. Jahangir, M. B. Riaz, M. J. Saeed, M. A. Sattar et al., ``Industrial applications of large language models,'' Scientific Reports, vol. 15, 2025, art. no. 13755

2025
[92]

Morales-Forero, S

A. Morales-Forero, S. Bassetto, and E. Coatanea, ``Toward safe ai,'' AI & SOCIETY, vol. 38, no. 2, pp. 685--696, 2023

2023
[93]

C. Reed, ``Meta fends off authors' us copyright lawsuit over ai,'' https://www.reuters.com/sustainability/boards-policy-regulation/meta-fends-off-authors-us-copyright-lawsuit-over-ai-2025-06-25/, 2025, accessed: 2025-07-04

2025
[94]

V. News, ``Judge backs anthropic: Ai training on legal books ruled fair use,'' https://www.vktr.com/ai-technology/judge-backs-anthropic-ai-training-on-legal-books-ruled-fair-use/, 2025, accessed: 2025-07-04

2025
[95]

R. Staff, ``Google's ai overviews hit by eu antitrust complaint from independent publishers,'' https://www.reuters.com/legal/litigation/googles-ai-overviews-hit-by-eu-antitrust-complaint-independent-publishers-2025-07-04/, 2025, accessed: 2025-07-04

2025
[96]

B. Gapp, . de Fauconberg, A. Bessis, H. Goble, and N. Obodoekwe, ``Digital finance in africa: Accelerating foundations for inclusive and sustainable local innovation,'' Environmental Sciences Proceedings, vol. 15, no. 1, 2022. [Online]. Available: https://www.mdpi.com/2673-493...

2022
[97]

Galaz, M

V. Galaz, M. A. Centeno, P. W. Callahan, A. Causevic, T. Patterson, I. Brass, S. Baum, D. Farber, J. Fischer, D. Garcia, T. McPhearson, D. Jimenez, B. King, P. Larcey, and K. Levy, ``Artificial intelligence, systemic risks, and sustainability,'' Technology in Society, vol. 67,...

2021
[98]

Göpfert, J

J. Göpfert, J. M. Weinand, P. Kuckertz, N. Pflugradt, and J. Linßen, ``Risks of ai-driven product development and strategies for their mitigation,'' 2025. [Online]. Available: https://arxiv.org/abs/2506.00047

2025 arXiv
[99]

Fenwick, E

M. Fenwick, E. P. Vermeulen, and M. Corrales, ``Business and regulatory responses to artificial intelligence: Dynamic regulation, innovation ecosystems and the strategic management of disruptive technology,'' Robotics, AI and the Future of Law, pp. 81--103, 2018

2018
[100]

Gra c a and L

P. Gra c a and L. M. Camarinha-Matos, ``Assessment of sustainable collaboration in collaborative business ecosystems,'' Computers, vol. 10, no. 12, p. 167, 2021

2021

Pith tools

Reviewed August 5, 2026 · model on record in the stance chip above.

[1] [1]

N. N. Mabaso, ``Evaluation of the implementation of the twin peaks financial sector regulatory model in south africa,'' Ph.D. dissertation, University of Pretoria, 2021. [Online]. Available: https://repository.up.ac.za/bitstreams/106d7444-3186-4884-ae40-53b858d8d611/download

2021

[2] [2]

Seiranian and A

M. Seiranian and A. Suzuki, ``Large language model parameter efficient fine-tuning for mathematical problem solving,'' Ph.D. dissertation, 07 2024

2024

[3] [3]

Adversarial AI in Insurance: Pervasiveness and Resilience

E. Luciano, M. Cattaneo, and R. Kenett, ``Adversarial ai in insurance: Pervasiveness and resilience,'' arXiv preprint arXiv:2301.07520, 2023

work page Pith review arXiv 2023

[4] [4]

63, 2025

Bank for International Settlements , ``Regulating ai in the financial sector: Recent developments and main themes,'' BIS FSI Insights Paper No. 63, 2025. [Online]. Available: https://www.bis.org/fsi/publ/insights63.pdf

2025

[5] [5]

C. E. Shannon, ``Prediction and entropy of printed english,'' The Bell System Technical Journal, vol. 30, no. 1, pp. 50--64, 1951

1951

[6] [6]

Hochreiter and J

S. Hochreiter and J. Schmidhuber, ``Long short-term memory,'' Neural Comput., vol. 9, no. 8, p. 1735–1780, 1997

1997

[7] [7]

Chung, C

J. Chung, C. Gulcehre, K. Cho, and Y. Bengio, ``Empirical evaluation of gated recurrent neural networks on sequence modeling,'' arXiv preprint arXiv:1412.3555, 2014

arXiv 2014

[8] [8]

J. D. Owens, M. Houston, D. Luebke, S. Green, J. E. Stone, and J. C. Phillips, ``Gpu computing,'' Proceedings of the IEEE, vol. 96, no. 5, pp. 879--899, 2008

2008