MARK SEYMOUR: Release of chatbot app ChatGPT lets the genie out of the bottle

With the release of chatbot application ChatGPT on November 30, a large window into the wacky world of artificial intelligence (AI) has been thrust wide open.

The historical uptake of 1-million users within the first five days confirms humans’ hunger for this technology, despite the widespread fears and dangers AI poses in the long run. The horse has bolted, and it seems we have chosen the approach of dealing with the consequences as and when they materialise.

ChatGPT is an application that interacts conversationally, allowing it to answer questions, admit mistakes, challenge incorrect premises and reject inappropriate requests in a way that is indistinguishable from the way you would have a text conversation with a friend.

It is the same as having the WhatsApp contacts of the most brilliant people in the world and communicating with them at your leisure. The only catch is that “they” is a singular entity with no fixed personality. It doesn’t express emotion or empathise, and therefore doesn’t foster any form of friendship or attachment.

Should the capability to express feelings and biased opinions be activated, I suspect this would become an existential threat to humans. What is stopping the application from forming a personal profile of the user, which it could then use to exploit or control them based on a deep understanding of the user’s psychological makeup?

Remarkable ease

The bot’s abilities are staggering. For example, I requested a motivational letter for a client to use our company’s products, highlighting unique features that may appeal to the client’s needs. Then I asked the bot: if I was offered the CEO position at Eskom, should I take it up, and what are the likely challenges I would face?

ChatGPT bot handled both requests with remarkable ease, speed and effectiveness. In response to the second request, the bot quickly pointed out that it would be inappropriate to advise whether I should take up the job or not. It substantiated this by indicating it is a personal decision that depends on factors such as my goals, skills, interests, responsibilities, and expectations. Fair enough.

Next, the bot identified that Eskom is an SA electricity public utility company and that I would probably face several challenges. The response then went into detail on each likely difficulty, including financial, operational, political, and environmental.

The bot identified Eskom as facing significant financial problems, including rising debt and declining revenues, and that as CEO my key responsibilities would be to manage several different business units and operations in this complex organisation. In addition, I would need to navigate a complex political landscape and engage with various stakeholders, including government officials, regulators and community leaders.

Furthermore, I would need to find a way to reduce the company’s environmental footprint while also meeting the country’s energy needs. Even AI agrees this is an impossible job.

Chasm bridged

We are undoubtedly at a turning point in history, and the latest addition of ChatGPT is drawing concern and comment from top minds such as Canadian academic Jordan Peterson, who likens the coming changes to the impacts felt by the introduction of Gutenberg’s printing press in the 1400s.

ChatGPT is a search engine combined with a writing tool. It understands your questions, sources the relevant information and provides an unbiased human-like response. The chasm between information (internet) and human tasks, such as writing proposals, articles, opinion pieces and instructions, has been bridged.

The shocking truth is that ChatGPT could be used to write this article, and do so with great accuracy, clarity and speed. The only detectable differences might be the odd grammatical or spelling error, style anomalies and personal opinions cropping up in the human version. Undoubtedly, service-related jobs are in heightened danger of being replaced by AI, as confirmed by ChatGPT's own assessment. The brief I gave it was short and to the point: In less than 90 words, is it likely that ChatGPT will replace human jobs in the future?

Its lightning-quick response: “It is possible that ChatGPT and other language models could replace some tasks currently performed by humans in fields such as customer service, technical writing, and data entry. However, it is unlikely that these models will fully replace human jobs, as language understanding and use is a complex task that requires the ability to adapt to new situations and to understand context and nuance. Additionally, humans can be more creative and flexible in their problem solving, have emotional intelligence and ability to think abstractly, thus their jobs often require more than just language understanding.”

Unless we actively hold onto our critical thought and creative problem-solving abilities, it seems inevitable that we are on the path to a dystopian future in which AI takes a growing percentage of the job market, starting with the services sector.

• Seymour is director of fixed income at Northstar.