Edition 84

Published
6 November 2026
Reading time
8 min
Contents
13 stories · 5 sectors

AI at work in South Africa, and what it still needs

South Africa's revised AI policy is likely to reach cabinet in March 2027, and Gleb Tsipursky argues it should test whether people can catch AI's errors. Mathews Phosa warned lawyers that citing AI-invented case law could end careers. An IITPSA survey finds demand for AI skills rising among employers, and Altron's CEO says its internal AI tools are not aimed at cutting jobs. Pick n Pay is rolling out an AI platform to more than 720 stores, and Discovery Insure says its AI driving coach is cutting risky driving. Further afield, Africa Finance Corporation's chief says AI cannot formalise Africa's informal economy without power and transport, researchers set out what farmers need from AI and blockchain, and TechCabal asks what "AI-powered" means for appliances. An EU court adviser says Gemini could breach press copyright rules, Google says schools that want no AI must leave Search off, Anthropic's IPO filing warns its progress could outrun its safety work, and Sam Altman's call to accept "bad things" has drawn criticism.

Every story headline links to its original source.

Policy & governance

  1. Sam Altman faces fierce backlash after saying society must accept ‘Bad Things’ for AI progress

    IOL (Independent Online)Policy

    OpenAI chief executive Sam Altman drew criticism last month after telling Politico that "the world should accept some bad things happening for the benefits of this technology", arguing that the good will far outweigh the harm. Michael Sherman, writing for IOL Tech, argues that Altman's stance treats human harm as acceptable collateral damage for a greater good, and that private tech executives should not decide how much risk the public absorbs.
  2. SA’s AI policy needs workers who can catch its mistakes

    Business Day / BusinessLIVEPolicy

    Gleb Tsipursky, a behavioural scientist and CEO of Disaster Avoidance Experts, argues that South Africa's revised AI policy, likely to reach cabinet in March 2027, should set a higher standard: organisations should show that human reviewers can actually detect and correct AI errors, not just promise oversight, starting with pilots in selected public services. Tsipursky proposes supervised error-detection training for junior workers and published correction metrics, warning that if entry-level checking tasks disappear, young workers lose the route to professional judgment.

Business & economy

  1. Pick n Pay turns to AI to speed up online orders

    TechCentralBusiness

    Pick n Pay is rolling out an AI-based store operations platform from Canadian company OrderGrid to more than 720 stores, after a pilot cut the average time staff took to pick online orders by about 30%. The platform manages stock levels in real time, forecasts demand and manages in-store tasks for the retailer's asap! online grocery service. Enrico Ferigolli, Pick n Pay's retail executive for omnichannel, said the tools will help teams pick orders more efficiently, particularly larger baskets.

    Also reported by ITWeb, Stuff South Africa

  2. Discovery uses AI to reduce risky driving

    ITWebBusiness

    Discovery Insure says its AI-driven Advanced Driving Instructor, launched on its app in January, has cut risky driving events among clients using it by 3% to 10%. The tool analyses individual driving patterns, such as speeding, harsh braking and phone use behind the wheel, and assigns personalised coaching goals. Chief analytics officer Michael Rodewijk says 68% of clients who achieve a goal keep the improved behaviour the following month, and that all analytics use only data clients have agreed to share.
  3. Tech won’t fix Africa’s informal economy without power and rails, says AFC CEO

    TechCabalBusiness

    Africa Finance Corporation chief executive Samaila Zubairu has warned that artificial intelligence and digital apps cannot formalise Africa's informal economy on their own. Speaking at the Africa Financial Industry Summit in Luanda, Zubairu argued that AI also poses real risks to labour markets and needs supporting elements such as reliable electricity, transport networks, tax systems and digital government services to lift productivity. With about 83% of Africa's employment in the informal sector, Zubairu said policymakers should plan whole regional ecosystems around infrastructure projects rather than digitise small enterprises in isolation.
  4. AI-powered appliances are coming to Africa. The infrastructure isn’t ready.

    TechCabalBusiness

    TechCabal's Kenn Abuya, after touring LG's model home in Seoul and factory in Busan, questions what appliance makers mean by "AI-powered": connected features can run on sensors and preset rules, and much of what is sold as smart goes missing without steady power and Wi-Fi. LG is pushing connected appliances across Africa, but World Bank and StatsSA figures Abuya cites show uneven electricity and internet access; South Africa's 90.2% electricity access and 82.1% household internet access suggest a market better prepared for connected products than Nigeria's. Likely early buyers are wealthier urban households, hotels and developers, and Abuya notes LG's factory-automation division poses a jobs-versus-investment trade-off for countries courting manufacturers, including South Africa.

Society & work

  1. Mathews Phosa urges lawyers to be wary of AI ‘hallucinations’

    Business Day / BusinessLIVESociety

    Mathews Phosa, former Mpumalanga premier, has urged lawyers to use AI responsibly when drafting court papers, warning that citing AI-invented case law could end careers and that lawyers who do it should be charged with unethical conduct. Speaking at LexisNexis' Future of Law Conference in Sandton, Phosa said the legal profession cannot avoid AI but must check everything it produces, and that regulation should also protect confidential information. His warning follows recent Gauteng and KwaZulu-Natal judgments in which lawyers cited non-existent case law, with judges reporting the practitioners to the Legal Practice Council.
  2. AI and blockchain are coming to African farms: 4 things must happen for farmers to benefit

    The Conversation AfricaSociety

    Olaoluwa Omoniyi Olarewaju of the University of KwaZulu-Natal and fellow researchers reviewed studies from 2005 to 2025 on AI and blockchain in farming and argue that the technology alone will not improve agriculture. Their paper says farmers need reliable data, tools designed for local conditions, clear rules on who owns farm data, and a clear economic benefit before adopting these systems. They warn that if smallholder farmers supply valuable data but have little control over it, digital agriculture could reproduce existing inequalities rather than remove them.

Education & skills

  1. AI wave creates new ICT skills demand in SA

    ITWebSponsoredEducation

    The 2026 IITPSA ICT Skills Survey, produced with Africa Analysis, finds that demand for AI-related skills is rising among South African employers rather than AI simply eroding ICT jobs. Lehlohonolo Mokenela of Africa Analysis says AI is mostly being used to automate routine tasks and support workers, with organisations facing skills shortages among the strongest adopters. The survey also finds growing demand for roles such as AI engineers, prompt engineers and AI governance officers, and it urges alternative learning pathways as traditional university and TVET pipelines fall behind.
  2. Google: Schools can keep search while skipping Gemini, but AI search is non-negotiable

    Ars Technica — AIEducation

    Google has clarified a new policy for schools using its education apps: from next year, its AI products stay off for logged-in student accounts until a school opts in, and Google counts Search among them. Schools can keep Search while skipping its Gemini chatbot, but AI cannot be removed from Search, so a school that wants no AI at all must leave Search off. The clarification follows criticism from educators and parents after early reporting suggested Google would block search for schools that declined AI.

Technology & infrastructure

  1. Anthropic “may in the future not identify” advanced AI model developments until they’ve happened

    Stuff South AfricaTechnology

    In an updated filing for its IPO, Anthropic says its growing use of AI to speed up its own research could push its models' capabilities forward faster than it can build matching safety measures, and that it has in the past not spotted such an acceleration until after it happened. The filing adds that Anthropic's internal estimates suggest the company may soon meet its Responsible Scaling Policy's definition of "dramatically accelerated progress attributable to AI", while noting that none of its current models has been designated as exceeding that threshold. Stuff's Brett Venter reads this as a hint of recursive self-improvement, where an AI model advances itself without human involvement, and calls the company's hints very vague.

More in this edition

  1. Altron hikes dividend following market share gains, AI Factory boost

    News24 / Fin24

    A year after launching its AI Factory, a specialised data centre for building and running AI tools that News24 describes as South Africa's first, Altron says more than 100 AI uses have been deployed across the group, with more than 500 employees using them daily. CEO Werner Kapp says no internal deployment is specifically aimed at cutting jobs, but also that customers can now do routine yet complex transactions faster and with far fewer people, and that marketing campaigns that took six months now take three to four weeks at 10% of the cost. Altron reported its best interim revenue growth in three years, with continuing revenue up 5.1% to R5.1 billion, and raised its interim dividend by 15%.

A “Sponsored” label marks content the original publisher was paid to run. The AI Factor carries no advertising.

Get the next edition

Confirm by email Published when there’s real news Unsubscribe anytime