ChatGPT has democratised AI, but like all new tech, comes with risks: analysts
By creating chatbot ChatGPT, artificial intelligence research and deployment company OpenAI has democratised AI, but, like all new technological developments, this carries the risk of challenges such as contextual errors, prejudice, information trustworthiness or credibility, analysts say.
The artificial intelligence language model has dominated news and social feeds since its release about three months ago, exploding in popularity due its ability to generate human-like text based on input received.
It is regarded as the fastest growing “app” in history, garnering over 100 million active users in two months, according to a UBS report using data from web and app analytics company Similarweb, which said over 13 million users accessed ChatGPT per day in January. This was more than double the previous month.
“In 20 years following the internet space, we cannot recall a faster ramp in a consumer internet app,” the firm was quoted as saying in the UBS report. For context, the most downloaded app of all time, TikTok, took nine months to reach 100 million users.
Microsoft, which already invested $1 billion in OpenAI in 2019, extended its investment in January with a multibillion-dollar deal which, according to Bloomberg, is worth $10 billion.
Following close behind, Google announced a ChatGPT rival called Bard on 6 February. Its experimental AI will be powered by its Language Model for Dialogue Applications (LaMDA) technology, unveiled two years ago.
Bard draws on information from the web to provide fresh, high-quality responses, Google chief executive Sundar Pichai said.
“Bard can be an outlet for creativity, and a launchpad for curiosity, helping you to explain new discoveries from Nasa’s James Webb Space Telescope to a nine-year-old, or learn more about the best strikers in football right now, and then get drills to build your skills,” Pichai said.
A day later, on 7 February, Microsoft announced its search engine Bing and browser Edge would be AI-powered using a next-gen OpenAI language more powerful than ChatGPT. The company said it aims to deliver better search, more complete answers, a new chat experience and the ability to generate content, calling the tool an “AI copilot for the web”.
“Until now, artificial intelligence could read and write, but could not understand the content. The new programmes like ChatGPT will make many office jobs more efficient by helping to write invoices or letters. This will change our world,” Microsoft co-founder Bill Gates told German business publication Handelsblatt, as reported by Reuters.
Earlier this month, ChatGPT introduced a subscription tier for $20 a month to access premium features such as availability when demand is high, faster response speeds, and priority access to new features. It is available in South Africa via a credit card payment through Stripe.
The technologies behind ChatGPT, such as Large Language Models (LLMs), Natural Language Understanding (NLU) and Neural Networks (NN) have been around for a while, but have only been accessible to a niche group, according to Natalie Raphil, founder of Robots Can Think.
“OpenAI, ChatGPT’s creator, democratised AI because it’s simple and accessible. People associate with its instant, personal value add and unfathomable opportunities. It responds to any prompt with comprehensive, seemingly plausible answers in startling speed. It is a new era of tech and AI,” Raphil, who holds a PhD in AI from Stanford University, told Mail & Guardian.
However, new eras of technology come with new problems, said Raphil, listing them as: plausible answers versus fact; contextual mistakes; prejudice; and information trustworthiness or credibility coming into question.
“Everyone is now an expert on every topic under the sun; it’s going to be even more difficult to separate the ones who talk the talk from those that genuinely are experts,” Raphil added.
The biggest challenge, she said, is that AI advancements are extremely rapid, meaning that governments, laws, regulations and education centres cannot keep up and are too slow to respond.
“The interconnectedness of technology is worrying for me, I recently saw a video of ChatGPT integrated to a 3D printer that hoisted a pen over a book … The end result was writing out school homework with one prompt,” Raphil said.
It is critical to understand that the AI used in ChatGPT — and indeed all generative AI — is backwards looking, Bronwyn Williams, a trend analyst at Flux Trends, told M&G.
“It can only recycle not create, which means it’s not creating ‘new’ things but rather finding common denominators or connections from what already exists,” Williams said.
“In many ways it is searching for the ‘right’ answer, aka the common denominator, aiming for the centre of the bell curve. And, secondly, this means there is plenty of room at the top for excellence and new creative ideas, and at the bottom it’s not good for mediocre ‘creatives’ and LinkedIn listicle writers.”
In response to Google’s Bard, a world stage was used to launch the difference between plausible and fact, Raphil said.
“The launch was an epic fail, which saw Google’s share price plummet considerably. Yes there will be rivals but I’m most looking forward to Microsoft’s Bing,” Raphil said. “Tech giants are in a race to launch the next best World Wide Web, but people are also in the race to have the world’s oil which is data.”
Big tech companies will race to claim power and control in the space, and Microsoft’s integration with ChatGPT is interesting in this regard, Williams said, adding: “One will likely win out — Google won the search race — but it is too early to tell who is furthest ahead here.”
“Other likely scenarios include different groups favouring different political biases — because unlike search, which gives the user options to choose from (divergence), these sorts of conversational AIs are designed to deliver a singular correct answer (convergence). Huge political and cultural power is at stake,” Williams said.
As a test, M&G asked ChatGPT what it thought of Bard, and the chatbox issued a reminder that as an AI language model, it didn’t have personal opinions or emotions.
“However, I can say that competition in the field of AI and language models is a good thing, as it drives innovation and helps advance the state of the art,” it said.
“Google’s Bard is just one of many language models developed by various companies and organisations, each with its own strengths and weaknesses. It’s always great to see new advancements in the field, and I look forward to seeing how Bard and other models will continue to evolve and improve.”
So, can chatGPT replace writers? Raphil said not yet. “Humans will be in the loop for a while to come. It surely does enhance the writer’s skill, allowing them to explore and expand their flair.”
It’s still very early days for AI and a glimpse of what is headed our way, Williams said:
“Generative everything — anything you imagine/want, whenever you want it — completely immersive satiation of mental desires is where we’re headed.”
“As Dan Faggella of Emerj AI says, the future of this technology is twofold: generative AI for power and control on the one hand (think business and propaganda); and generative AI for distraction (think interactive virtual sexbot girlfriend) on the other,” Williams added.
Published here: https://mg.co.za/sci-tech/2023-02-14-chatgpt-has-democratised-ai-but-like-all-new-tech-comes-with-risks-analysts/
Recharged is an independent site that focuses on technology, electric vehicles, and the digital life by Nafisa Akabor. Drawing from her 16-year tech journalism career, expect news, reviews, how-tos, comparisons, and practical uses of tech that are easy to digest. email@example.com