Is ChatGPT hyped or simply disruptive?
Imagine writing that tricky essay for your English homework. It's late at night and you must hand it in first thing in the morning. Simply type a few keywords into a chat window and hey presto, that urgently-needed essay or advertising copy and even a computer program appears in seconds thanks to the AI behind ChatGPT ("Generative Pretrained Transformer). Developed by the U.S. company OpenAI, ChatGPT’s results are uncannily similar to those of a human author. The British mathematician, Alan Turing, first came up with a definition for artificial intelligence in 1950. If a computer’s response cannot be distinguished from that of a human being, then a computer is intelligent. That begs the question of whether ChatGPT really is artificial intelligence. Hamburg News spoke to Werner Bogula, Digital Enabler at the Artificial Intelligence Center (ARIC), the main point of contact and co-ordination point for artificial intelligence in Hamburg.
Hamburg News: Mr. Bogula, how intelligent is ChatGPT?
Werner Bogula: ChatGPT is not intelligent in the sense that it understands the content of a question, but rather it considers "input" an incomplete text that it completes....similar to predictive text on a mobile phone, but on a much bigger scale. To this end, ChatGPT was trained with huge amounts of text - more than 500 billion words - and to make connections between words, sentences and facts. To train this model, large parts of the internet were read in and the correlations between the words in different languages were calculated as a statistical model in a huge computer centre. ChatGPT is unique in that was trained in talks with human beings and thus answers plausibly and like a person.
Hamburg News: This special feature is attracting a lot of attention at the moment. Is it justified?
Werner Bogula: ChatGPT is indeed on everyone's lips at the moment and for good reason. A large part of our social and economic interaction in the information society is per text. Having AI that can generate such texts in appropriate contexts is highly relevant in both social and economic terms. The ChatGPT-generated user texts, e.g., blurbs, advertisements for the internet or brainstorming for new product ideas give professional copwriters in advertising agencies, publishers and on news websites a powerful tool for creating routine texts quickly and cheaply. However, such a powerful tool also casts doubt on existing structures. If texts that are normally written by professional copywriters can be generated at the press of a button, that puts pressure on existing job structures. It remains to be seen whether such jobs simply disappear or become more productive. But this is only the beginning. We at ARIC are addressing how to regulate access to such enabling technology and its results. Our keyword is "Responsible AI".
Hamburg News: How important is this enabling technology for Hamburg's economy?
Werner Bogula: Apart from the knowledge already fed into the model, ChatGPT can also use its own knowledge. That brings us to business. Corporate knowledge management is becoming an increasingly competitive factor. A chat bot that has been retrained can make this knowledge available to all the employees. Two aspects are important for commerce in Hamburg. On the one hand, AI companies and start-ups can adapt this new new technology so that all kinds of businesses can use it themselves. Take for instance, Moin.ai, which has developed over 100 chatbots for businesses and organisations. Neuroflash is an another example. This extremely innovative start-up has developed an easy-to-use interface that can generate over 100 types of text - from ads, social media posts to entire blog posts.
On the other hand, companies can benefit from such start-ups. We at ARIC are in talks with several companies that want to break down their proprietary knowledge better and to make disseminated knowledge available to their customers and employees through a standard chat interface.
Hamburg News: Is ChatGPT simply a hype or a technical revolution?
Werner Bogula: Of course it is hype. More than a million people signed up for the technology platform in five days when ChatGPT was released last November. Students are suddenly astonishing their lecturers with interesting essays. But this technology is revolutionising the way we work with linguistic knowledge. The above examples highlight the situation from different business perspectives. And that is only the beginning. Decades of research and development in computational linguistics have gone into ChatGPT.
Hamburg News: Mr. Bogula, many thanks for the interesting talk.
Interview by Yvonne Scheller
P.S.: Asked how revolutionary ChatGPT considers itself, the bot was diplomatic...it depends on the perspective.
Werner Bogula joined ARIC in mid 2022 and helps companies and organisations to launch new technologies in a socially-responsible manner as part of the AI network Regional Future Centres. After founding a start-up and working as a developer and technology consultant in Silicon Valley, South Africa, India and Vietnam, Bogula, who is a qualified computer linguist and communication scientist, has vast experience using AI and dealing with its repercussions on society and commerce.
ARIC will be holding a free online workshop with Werner Bogula focusing on whether ChatGPT is the ultimate disruption for artificial intelligence. Registrations are being accepted on: www.aric-hamburg.de/veranstaltungen