ChatGPT one year on: How has it affected the way we work?

ChatGPT has made it much simpler to manage complex tasks like crafting and editing articles or creating nutrition plans. The chatbot can deliver results in seconds. ChatGPT can even generate detailed construction plans or computer programs in a matter of minutes — assignments that take architects and coders many weeks of hard work.

When US tech company OpenAI launched ChatGPT one year ago on 30 November 2022, it sparked a veritable revolution. The chatbot uses machine learning and was trained on a massive body of text data. It has since been programmed to search the internet for pictures and documents after receiving a query.

Millions of people started using ChatGPT immediately after its launch, according to OpenAI. Now, one year later, even more have made use of the platform.

In August alone, the ChatGPT website received some 1.43 billion visits, according to Similarweb, an online traffic analyst. These days, however, many users tap into ChatGPT's neural network via other apps instead of through its dedicated website. The total number of ChatGPT users is therefore likely to be significantly higher.

ChatGPT will likely have a direct impact on the world of work. Investment bank Goldman Sachs for instance estimates that artificial intelligence (AI) systems such as ChatGPT could lead to the automation of up to 300 million jobs worldwide. Academic and creative professions would be most affected.

And while this may sound like bad news, it does not have to be. While Goldman Sachs says the "impact of AI on the labor market is likely to be significant, most jobs and industries are only partially exposed to automation and are thus more likely to be complemented rather than substituted by AI."

Daniel, a software developer employed by a large corporation, agrees with this optimistic assessment. He does not want to reveal his full name, however, citing competition among tech companies.

"You can't give a detailed programming assignment [to ChatGPT] and expect perfect results," Daniel told DW. "It works best for me if I take ChatGPT's answers as a suggestion and then continue to build on that to create something better in dialogue with the AI platform."

And so far, ChatGPT results are not always reliable. Even if they can look convincing at first glance, they sometimes amount to utter nonsense. OpenAI has said it plans to iron out such AI "hallucinations" in the next one to two years.

Some tech experts, however, doubt such "hallucinations" will ever disappear. This means if individuals use ChatGPT to answer very important questions, they should carefully check the results.

German international broadcaster DW is also experimenting with AI. But "generative AI will not replace the work of our journalists," according to DW editor-in-chief Manuela Kasper-Claridge.

Universities are somewhat concerned about ChatGPT and the risk of plagiarism. German universities, for example, have not yet established a set of common rules for dealing with ChatGPT-generated material.

"While there is consensus that the use of such tools should not be banned, a critical approach should be encouraged," Martin Wan of the German Rector's Conference's digitalisation unit told DW. "In terms of academic integrity, this includes being transparent about the use of such tools and not passing off ChatGPT-generated texts as one's own work."

Wan says while one may use such system to "compile the latest research findings" one should bear in mind they are still "hallucinating" and should be cautious about AI-generated content.

It therefore seems unlikely that ChatGPT could be used to commit large-scale plagiarism. "Even when AI generators provide something akin to sources, this is rarely like a scientific bibliography," Wan told DW. "You realize that relatively quickly once you get into a subject."

But ChatGPT can also be utilised for studying. Dom, a student, told DW he primarily uses the platform to better understand complex topics. "It's like having someone sitting next to you that you can ask. I know it's not entirely accurate, but it gives me a point of reference," Dom said. "I wouldn't have passed the last exam without ChatGPT."

Teresa, another student, told DW she uses ChatGPT too." But I'm not sure what you are allowed to do with ChatGPT," she told DW. "So far, I've mainly used ChatGPT to get a quick overview [over different subjects]."

Other students told DW they are still cautious about using the tool even though it was introduced an entire year ago. "I don't trust it, there's no guarantee that it's right," Mariya, another student, told DW.

It is hard to say whether one should remain wary of ChatGPT or fully embrace the technology. But rumours surrounding recent changes in OpenAI's leadership, including the firing and rehiring of Sam Altman, suggest that the company is already working on a new AI system that could overshadow ChatGPT.

This article was translated from German.