AI writing tool ChatGPT is taking the world by storm. What exactly is it? What’s its potential, for good or ill? BV editor HAL WILLIAMS investigates, while pictures editor RICHARD THOMAS puts questions to the bot itself …
A TECH disruptor is on the loose; it’s lean, mean, growing like a virus — and coming for all things you hold dear, songs, news reports and literature among them.
Or maybe it’s here to make our lives easier, more colourful, more creatively fulfilled; no one knows for sure, just yet.
Joining deepfakes, Terminator and robotic lovers in our growing cabinet of technological horrors-slash-wonders is a chatbot that’s easily accessible — and free, during a feedback period by its developer, OpenAI. It can pen articles or epistles built with words, phrases and even inflections plagiarised from countless, often unidentifiable sources. It can plunder the finest minds on the web and package their wisdom. You can present it as your own.
Meet ChatGPT.
The chatbot was launched by OpenAI in November last year. In the few months since, it has become a media sensation.
It’s built on top of OpenAI’s GPT-3 family of large language models and is fine-tuned with supervised and reinforcement learning. Already — and the thing’s basically a prototype — it can skim through the toughest academic exams as though they were finger-painting exercises for toddlers, causing some universities to consider returning to a pen-and-paper format for finals.
A sensation, certainly. But is it all it’s cracked up to be, in any sense? Behavioural scientist Nick Hobson was recently quoted as saying: “The robot is capable of answering all of your questions and prompts like an intelligent human would. The model can be fine-tuned for specific tasks, such as language translation, summarization, and question answering.”
He goes on to list several ways ChatGPT can be made to work for you and increase productivity. More common is speculation about how many jobs will be lost — in media and comms, for starters. Are poets doomed to never out-prose a bot? Is the soul going to be sucked out of society, one literary gem at a time?
Not so fast, says Yann LeCun, Meta’s chief AI scientist. He has described ChatGPT as “not particularly innovative” in terms of underlying techniques. “It’s nothing revolutionary, although that’s the way it’s perceived in the public,” he said. “It’s just that, you know, it’s well put together, it’s nicely done.”
“Nicely done?” Is LeCun damning this earth-shattering tech with faint praise? Is this a case of, as a Meta man, “he would say that, wouldn’t he?” Or is he, perhaps, secretly concerned about the moral and ethical aspects of ChatGPT, and emerging similar systems? On reflection, go for option two.
But ethics is of concern. Intellectual property, copyright, plagiarism, theft… all sort of ugly words start popping up. What checks and balances are in place — as ChatGPT shoots over the horizon like a supercharged bottle rocket — and how many more will be needed?
Lance Eliot, of Forbes magazine, was recently quoted as saying there are ongoing efforts “to imbue Ethical AI principles” into the development and fielding of AI apps. “A growing contingent of concerned and erstwhile AI ethicists are trying to ensure that efforts to devise and adopt AI takes into account a view of doing AI For Good and averting AI For Bad,” he said.
“Likewise, there are proposed new AI laws that are being bandied around as potential solutions to keep AI endeavours from going amok on human rights and the like.”
Effective and ethical or not, EY’s global chief technology officer Nicola Morini Bianzino tentatively questions the validity of it all. “Right now, there is no ‘killer’ use case for using ChatGPT in the enterprise,” he said in a recent interview, “that is, one that will have an enormous impact on the top and the bottom lines.”
Bianzino then goes on to predict “an explosion of experimentation” in the next six to 12 months, especially once companies are able to build on top of ChatGPT using OpenAI’s API. “And the killer use-case that emerges could be around generative AI’s impact on knowledge management.”
Israel’s president, Isaac Herzog, admitted in a video greeting for Cybertech Global Tel Aviv 2023 that the opening part of his speech had been written by ChatGPT. Democrat congressman Jake Auchincloss broke the Congress chatbot cherry by reading a two-paragraph speech generated by ChatGPT AI bot on the floor of the House of Representatives.
The tech is, of course, about to be copied — and challenged — by competitors. Google has announced it’s launching a rival to ChatGPT: a bot called Bard. It’s being tested as we write, and will be rolled out to the public within the month, says Google. Bard is built on Google’s large-language model Lamda, famously described by one engineer as possibly being sentient.
And therein, for many, lie future concerns. The insanely rapid progress, and ChatGPT’s instantaneous functionality, have caught humanity on the hop. Yes, it can do all this stuff — right now. But, um, what do we want it to do? And, if there are things we want it to not do, how easy is it to restrain it?
One thing that’s become clear while researching this article is that AI, and its developers, have the digital pedal to the virtual metal. Everything’s flashing past so quickly it’s impossible to keep up — and ChatGPT and its ilk are clear for take-off.
A chat with ChatGPT…
BV’s RICHARD THOMAS delves into OpenAI’s world and puts questions to Chat GPT…
Do you think you’re a threat to human creativity?
As an AI language model, I don’t have personal thoughts or opinions. My purpose is to assist and provide information, not replace human creativity. In fact, AI has the potential to enhance and augment human creativity by providing new tools and resources for creative expression and problem solving. Ultimately, the impact of AI on human creativity depends on how it is developed and used.
What impact will artificial intelligence have on human creativity in the fields of literature, art and journalism? Explain to an international business, senior management audience
Artificial Intelligence has the potential to have a significant impact on the fields of literature, art, and journalism. In literature, AI can assist writers in generating new story ideas, character development, and plot structure, among other things. However, it is unlikely that AI will completely replace human writers in the creation of truly unique and compelling works of fiction.
In the field of art, AI has already been used to generate new forms of digital and interactive art, such as generative art, which is created using algorithms. AI can also assist artists in tasks such as image manipulation, color correction, and composition, freeing up more time for the artist to focus on the creative process.
In journalism, AI can be used to perform tasks such as data analysis, fact-checking, and content generation. This can help journalists produce more accurate and informative articles, while freeing up more time for research, investigative reporting, and other tasks that require human creativity.
It’s important to note that while AI has the potential to augment human creativity in these fields, it cannot replace the unique qualities that come from human experience, imagination, and emotional intelligence. Ultimately, AI will likely have the greatest impact in these fields when it is used as a tool to support and enhance human creativity, rather than as a replacement for it.
In answering simple questions, how do you differ from Google? Are your results superior?
As a language model developed by OpenAI, I am trained to provide information and answer questions based on patterns in the text I was trained on. Google, on the other hand, is a search engine that uses algorithms to search the internet and return the most relevant results based on your query.
In terms of providing information and answering questions, I can provide more in-depth and specific answers compared to Google, as I have been specifically trained to generate human-like text. However, the information I provide may not be as up-to-date as Google’s search results, as my training data only goes up until 2021. Additionally, Google has access to a much larger volume of information, so it may be better suited for more complex or current queries.
