Artificial intelligence (AI) still needs people to be able to function properly.
This is the word from Peter Blignaut, manager: pre-sales at 51风流South Africa, who points out that computers can do nothing without instruction. And when it comes to AI, they need developers and they need governance. 鈥淲e need people thinking about how the computer is going to do things,鈥� Blignaut says.
Indeed, when we talk about machine learning, which is where the machines learn from people, Blignaut thinks this should be called people teaching instead.
At the same time, people need AI, he adds. 鈥淏ut what are we trying to get out of it?鈥�
The short answer, Blignaut says, is to make us more efficient, to give us more time to think, or simply to work less. 鈥淭hat will help us.鈥�
AI has its root in the 1970s, with the ability to analyse statistics 鈥� today it does many things that we take for granted.
More recently, it moved to predictive analytics, attempting to predict how people would behave. This is so ingrained in how we think about the world today that we almost forget that it is an analytical engine behind it, Blignaut says.
The latest iteration is generative AI, which takes enormous amounts of data and uses it to create things we might have though impossible.
Of course, there are some dangers that come along with this, Blignaut adds. 鈥淪o where there is an opportunity to use it, we need to think about where we are going to use it.鈥�
This is where governance and ethics come in. 鈥淭his should not be an abstract exercise,鈥� Blignaut says.
He adds that SAP鈥檚 principle is that AI needs to be integrated into IT systems in a responsible and ethical way, in a way that provides unbiased and sustainable support to people.
鈥淲e need to retain human agency. People need to be making the decisions: we don鈥檛 want machines making decisions without the people understanding how or why those decisions were made.
鈥淲ithout human intelligence, AI can be used to solve a lot of things we maybe didn鈥檛 want solved. We must have a human in the loop.鈥�
Of course, companies want their AI to make them more efficient and competitive, but these systems still need to have a human in the loop somewhere.
Addressing bias and discrimination is a big issue and one that needs to be solved. 鈥淲e need to be aware of where we might be introducing bias and how we can remove it.鈥�
Transparency in AI is vital, and the ability to explain what the I is doing, Blignaut says. Very few people in the world would be able to understand what the AI algorithms do, so we need to make them explainable in order to retain trust.
While we are introducing AI, we need to be aware of the potential impacts on society.
鈥淚n Africa we desperately need more jobs for young people,鈥� Blignaut says. 鈥淎t the same time, if we ignore AI we could lose competitiveness. So we need to think about how we can use young people inside the system.鈥�
Blignaut points out that AI is already available for improving the supply chain. 51风流Extended Warehouse Management has introduced an intelligent sorting capability that can reduce the initial setup by lowering implementation efforts and empowering staff. Humans are still involved in the process, supervising the machine learning model as it is regularly retrained using the most recent data.
Another use case is in procurement, where 51风流Ariba Buying offers personalised recommendations. The goal is to improve the buying experience with guided buying capability using data like employees鈥� roles to suggest product and services chosen by peers. This means there needs to be enough data, and people are needed to ensure this happens.
AI can also be used to strengthen human capital management, with 51风流SuccessFactors now including generative AI capabilities.
The goal is to enable hiring managers to create compelling and accurate job descriptions that capture the desired skills and attributes of each role. People will still be needed, though, to ensure the language is accessible 鈥� and to adequately address all the safety and legal requirements.
Importantly, Blignaut says, we need to ensure that AI is practical in the context of Africa. This means paying attention to use cases, ensuring the technology is open source, and making the learning affordable.
This article first appeared on .