Sentient AI: have we entered The Matrix?

post-title

Blockchain guru Professor Steven Sidley gave CIO South Africa Summit attendees a look into its history.



During a CIO South Africa Summit in Cape Town on 20 April, Steven explained what it could do for business, but also its impact on society as a whole. But where did it all begin, and why is it important?

According to Steven, AI started to surface around the 1980s. Not many knew what it was actually about. “We’ve seen AI in movies like The Matrix, The Bladerunner, and games like AlphaGo Zero. Therefore, the thinking about artificial life and artificial intelligence has been embedded in our culture, politics, and technology for thousands of years and it has been slowly creeping up on us,” he said. AI’s most talked about rockstar, ChatGTP, launched in November last year.

According to Steven, it’s moving at lightning speed. ChatGPT enters the scene, it’s creating a buzz, it’s still a baby, and three months later, enter GPT-4, which is way more powerful. “Therefore, the argument about ethics and sentients being a thing of the future is not in the future, it’s coming at us exponentially accelerating speed,” he said.

Steven called ChatGPT “the fastest uptake of software in history,” racking up a million users in its first week, and one hundred million users in a month. People who had never heard about artificial intelligence before were suddenly talking about it everyday. “It does one simple thing, it is very good at avoiding plagiarism, it will respond to any prompt about a subject, it sometimes sees things that aren’t real, it cannot explain how it responds to questions, and it’s ability to understand and respond to 26 languages is unsettling,” he added.

“It thinks probabilistically and not determinedly, where it sometimes chooses something on the far end of probability (irrationally or unlikely), and that’s probably the reason why it hallucinates sometimes,” he joked.

Surprisingly enough, Steven highlighted, is that its own creators cannot explain how it works, which is something critical to note because no other piece of technology has been created where you cannot explain how it works. That’s why Italy recently banned ChatGPT because its law says that if you sell something to a consumer, you at least have to explain how it works.

“There are no handrails or precedents to predict the impact of an autonomously and exponentially improving technology.”

“AI is entirely different from all the other transformative technologies that humanity has ever created. It is different because it is the only piece of technology that has been created that learns. That, fundamentally and profoundly, distinguishes it from the other technologies,” Steven noted.

He highlighted that other technologies have been improved in small increments and, sometimes, in large increments. However, none of them have the ability to learn. Moreover, in principle, AI learns autonomously and without our input, “‘exponentially’ meaning that we have no idea what it’s going to look like next week, never mind in the next 10 years.”

Steven explained that there was a great split in the year 2000. “Prior to 2000, people were trying to do AI by trying to attach semantics to things – heuristics, essentially. However, machine learning (ML) decided to take a different approach of gathering a whole bunch of data and statistics instead. ML is compute intensive, speed intensive, and storage attentive, and the mathematics of that statistical engine started to get very sophisticated – that’s when it took off.”

On software development, Steven used GitHub Copilot, a cloud-based artificial intelligence tool developed by GitHub and OpenAI, as an example for how powerful this technology has become. “Replit, for instance, is writing software at senior software development level!”

A lot of experimentation has been done with many of these AI generated codes, he added. For example, people have given AI tools very complex pieces of code and asked them to find a security vulnerability in them – and they were able to do so. The danger here, says Steven, is that someone will eventually take that AI and breach or collapse financial systems to create dangerous weapons.

“This is going to happen because AI is a powerful weapon. We have to stop, but we can’t. There are people out there who don’t have our worldviews,” he noted.

He said he didn’t see it replacing people and their jobs. But one attendee asked: “If it is able to make us more productive, then the issue of being able to do more with less, means that you would need less people in the physical sense, which means that you need less people to do more work, right?”

“My confidence is that AI will make workers more productive, in other words, move them up the productivity chains in ways they wouldn’t be able to do in our current environment. The productivity issue is very important, because it takes humanity up to another level of doing stuff that humanity thinks is important, and leaves the stuff that can be done better and faster to technology,” Steven concluded.

Related articles

How and where will the future CIO work?

What is the workforce of the future? Who will be doing the work? And where will you be doing work? During a discussion with Eskom CIO Faith Burn at the 2024 CIO Day, Investec CIO Shabhana Thaver discussed the role of IT in shaping future work.

Warren Hero joins SARS as new CDO

The 2023 CIO Awards winner will be responsible for designing the South African Revenue Services’ business model for antifragile digital transformation.

Top