“I’m not a believer in free artificial intelligence (AI) at all costs; I do believe there has to be regulation when it comes to the transparency of AI.”
So said Arthur Goldstuck, author, ICT commentator and World Wide Worx CEO, at the launch of his latest book – “The Hitchhiker’s Guide to AI: A handbook for all”.
The book was launched last night at Exclusive Books Sandton, in conversation with 702 presenter John Perlman.
Published by Pan Macmillan South Africa, The Hitchhiker's Guide to AI is Goldstuck’s 20th book and offers an overview of the past, present and future of AI. It is aimed at beginners, as well as those who consider themselves experienced or skilled at using AI.
It draws on years of direct access to global and regional leaders in using AI, across Africa, the Middle East, North America, Europe and Asia. It provides perspectives on generative AI, as well as practical advice for using it, reads the synopsis.
In the wake of the AI race, there is heightened activity and a shift in focus by regulators, to ensure the technology aligns with governance principles.
“I do believe that one of the regulations that will begin emerging as governments learn how to grapple with this…the language models that are being used, the way they are trained and the data that goes into them is going to have to be transparent,” he commented.
Goldstuck pointed out that this week Google announced Gemini, its latest AI model, and promised transparency in how the AI is trained.
“I think they have recognised the regulators are coming for them, if they don’t do it. I suspect we’ll see more and more of that kind of regulation, and I don’t think we’ll see too much objection from the AI industry because they understand they also have to earn the trust of the consumer and governments.”
While AI has been around for a few years, the past year saw big tech players – including Microsoft, xAI, Meta and Amazon – make large investments into AI technology as they look to improve their AI tools.
OpenAI, the company behind ChatGPT, recently released an updated version of the generative AI tool. The update includes new features, such as prompt examples at the beginning of a new chat, suggested replies, the ability to ask ChatGPT to analyse the data of multiple files, the capability to stay logged in for longer and keyboard shortcuts for faster use of the AI chatbot.
X Corp CEO Elon Musk launched his own AI firm in July, called xAI, which he said is going to challenge ChatGPT.
Meta and Microsoft in July partnered to open their AI model Llama 2 for commercial use. Llama 2 is a free-to-use model that enables developers and organisations to build generative AI-powered tools and experiences.
Goldstuck’s book also addresses the issue of bias and how AI can reinforce it, noted Perlman, referencing the author’s observations that “machines learn from the data with which they are fed”.
Providing context to this, Goldstuck explained: “One of the companies I write about in the book is Amazon, particularly because I attend the Amazon Web Services re:Invent conference, where they announce all their latest innovations. For the last six years, [Amazon] heavily focused on AI. Back in 2017, they unveiled 22 new AI products.
“I’m obviously a fan of these companies and what they are doing, but I’m not blind to the realities and the mistakes they make.
“For example, a couple of years before, Amazon was using AI to assist in recruitment, assessing CVs and selecting them to pass on for hiring new staff. Someone discovered there was a massive selection of males through this process. When examined, they discovered an AI tool had been fed with the CVs of hires from the first 10 years or so of Amazon, which were primarily male.
“There was a massive male bias and anyone who was identified as a woman or female by the AI was scored down as a result of the bias of the data that had been fed into it. That’s not a one-off example, it’s a one-off for Amazon because it woke them up them up in a very big way that they actually scrapped the software and eventually closed down the division that had come up with that software.”
Goldstuck pointed out that other companies took many years to learn that same lesson, especially in the US banking industry, with AI systems being severely biased against people of colour and those living in lower socio-economic communities.
“The banks’ AI systems scored people according to who had been successful mortgage owners in the past, and those tended to be affluent white people who lived in more affluent areas.
“These are examples of how the data that is fed into AI can result in severe bias, and it’s one of the great warnings against blindly using AI.”
While there’s been a proliferation of AI tools over the course of the last year or two, Goldstuck says some will probably die or be absorbed into other tools. “What will certainly happen is that every one of them will become more powerful and morph into more advanced tools, possibly different names and elaborate names.
“The one thing that will certainly happen is that there will be a plethora of new tools that will be as good, if not better than the ones that are currently out there.”
The Hitchhiker’s Guide to AI is available at select bookstores nation-wide.