Beyond the hype, AI offers practical solutions for Africa's unique challenges and is set to transform industries, the world of work and people’s lives.
This is according to experts at Hewlett Packard Enterprise (HPE) and its partners, who were addressing delegates at the HPE Innovation Day, in Johannesburg, this week.
President Ntuli, MD of Hewlett Packard Enterprise South Africa, said: “We have an opportunity to challenge the status quo in Africa, leapfrogging other continents and learning from their successes and mistakes to develop AI in Africa – for Africa. Our customers, especially in South Africa, see AI as both a challenge and an opportunity,” he said.
Ntuli highlighted the need for the appropriate infrastructure and responsible data governance to maximise the AI opportunity for SA.
He noted that HPE had a long history of harnessing AI, with established centres of excellence. “We provided AI capabilities way before the advent of generative AI,” he said. Today, HPE works in partnership with companies such as AMD, Cohesity, Microsoft and Red Hat to help customers navigate complexities and operationalise AI.
Operationalising AI
John Carter, VP, server product, quality and technical pursuit at Hewlett Packard Enterprise, said: “Only 33% of customers have an enterprise AI strategy, but I guarantee that 100% of them are talking about it. Having a strategy is the first step towards successful AI. Only 10% of AI projects actually make it to production; however, this is a good sign. With smaller projects, you can fail fast and learn from them. And because it takes seven months or more for an AI project to go from pilot to operationalisation, it is important to start early.”
He pointed out: “AI isn’t magic – it’s a standard workload, but these new technologies are extremely power hungry, so you have to create efficiencies in your data centres to run these workloads.”
Carter outlined various types of AI, including advanced analytics, small language models, large language models (LLMs), agentic AI and artificial general intelligence, and noted that few organisations and use cases require full LLMs and the resources they need. “You can start with much less and be very successful. You don’t need to be able to do everything like Chat GPT can – small language models are well suited for specific sets of data and specific tasks within organisations,” he noted.
He said AI was proving highly successful in highly regulated sectors that already had access to large volumes of quality data – such as banking, insurance and healthcare, and was also being deployed successfully in sectors such as retail, telecoms, media and entertainment, manufacturing, government and energy.
When looking at AI use cases, look for repeatable actions with many possible outcomes, he said, explaining that AI differs from automation in that it addresses repeatable actions with limited outcomes.
Carter advised that the keys to successful AI adoption included having a narrow range of well-scoped tasks, abundant training data, secure and structured data, a ‘humans-in-the-loop’ model, seamless integration and the right partners.
“This stuff is hard. It’s complicated. Having the right partners like HPE and our partners simplifies it,” he said. “Don’t get paralysed by the choices – start small, fail small and move forward,” he said.
Outlining HPE’s AI capabilities, Carter said HPE private cloud AI offered AI models, software, infrastructure and services. HPE’s new Proliant servers also catered for a range of AI use cases – from the new DL145 to the DL380a Gen 12 and the DL 384 Gen 12.
Securing AI
Chris Klosterman, cyber resiliency technical sales specialist at Cohesity – a company specialising in AI-powered data security and protection – said the top data challenges in AI include large volumes of data to protect, securing and recovering data, and extracting insights from data.
“Gaia was the first to offer retrieval augmented generation (RAG) on secondary data, allowing organisations to combine generative AI with their own, trusted internal data,” he said.
AI for Africa
A customer panel discussion chaired by Arthur Goldstuck, founder and CEO of World Wide Worx and author of "The Hitchhiker's Guide to AI", considered issues such as AI’s impact on Africa, and governance and ethics around AI.
On the panel were Carter; Johan Steyn, AI and automation consultant; Warren Hero, head of Segment, Product and Process at SARS; Bramley Maetsa, digital innovation enablement lead at Sasol; and Francois Swanepoel, IT operations manager: cyber security, ITSM and infrastructure at King Price.
Delegates also learned about leveraging hybrid cloud to make AI real from Dean Erasmus, chief data officer at Microsoft; AI-powered networking by HPE Aruba Networking from Mandy Duncan, Aruba country manager at Hewlett Packard Enterprise; Red Hat's vision for AI from Bruce Buskansky, app platform specialist at Red Hat; and management in a hybrid world with AI-infused observability from Saleh Al-Nemer, UKIMEA enterprise architect, distinguished technologist at Hewlett Packard Enterprise.
Share