Saphila: AI needs people

Computers are essentially inert without instruction, and it is the role of data scientists to provide them with the guidance they need, the Saphila 2023 conference at Sun City heard this week.

Saphila is a biennial conference for the SAP user group community, hosted by the African SAP User Group (AFSUG) in partnership with SAP.

“A computer does nothing without instruction… We need people thinking about what the computer is going to do,” said Peter Blignaut, manager of pre-sales at SAP South Africa during a keynote address. “Machines don’t learn from other machines, people teach them”.

He says that AI, particularly machine learning, which enables computers to learn from data and improve their performance, is fundamentally a product of human teaching rather than machines teaching themselves. 

Blignaut believes that AI has the potential to enhance our efficiency and grant us more time for meaningful tasks. By delegating repetitive and mundane tasks to computers, it will be possible to work more efficiently and devote attention to endeavours that require human ingenuity. However, he argues that the implementation of AI should not be solely driven by profit motives. Instead, it should aim to make us more effective as individuals, enabling us to become better versions of ourselves.

Involving people in the development and application of AI is crucial, and Blignaut stresses the importance of governance and ethics in this domain. The technology requires thoughtful deliberation and a framework for responsible decision-making. However, the practicality of implementing such governance measures in countries like South Africa .

An example of everyday predictive AI is Netflix and its personalised content algorithms that recommends content to its user based on their previously watched content. These algorithms are far too complex for manual execution, making it necessary for computers to take over this task. This integration of AI into our daily lives underscores its potential to enhance our experiences and streamline various processes. 

Newer generative Ais, such as Chat GPT and Midjourney, hold great promise for the future. By feeding vast amounts of data into these systems, one can generate diverse outputs, ranging from art to essays. However, Blignaut warns of the dangers associated with generative AI, particularly the potential for individuals to create fake content. The responsible use of this technology requires careful consideration of where and how it is employed, weighing the benefits against potential risks.

Blignaut says it is crucial to maintain human agency in AI, ensuring that decisions made by machines are comprehensible to humans, so that a person understands the decision-making process and can verify its logicality. This understanding fosters trust between the parties involved, further reinforcing the argument for promoting AI education. 

If people are unfamiliar with new technologies, they tend to avoid them, which is why it is essential to provide explanations about what algorithms are doing.  As AI continues to evolve, it is important to approach its implementation thoughtfully, harnessing its potential to enhance lives while safeguarding against potential pitfalls.