Microsoft has announced the availability of new versions of Bing and Microsoft Edge powered by an update to ChatGPT, the chatbot recently announced to replace Google. Initial feedback is available: Bing showed 2.5 billion people as an answer to the question of how many people live on the planet Mars. The situation begs the question of whether the rise of artificial intelligence will instead make people work harder to combat misinformation.
The chart follows a recent warning from Apple co-founder Steve Wozniak: The problem is that it does good things for us, but it can make terrible mistakes if it doesn’t know what humanity is.
In fact, it is a repetition of the opinion of some teachers. Ethan Mollick is one of them and has opted for an open usage policy of the chatbot ChatGPT. The reason: The use of artificial intelligence is an emerging skill. Nevertheless, he uses the latter to specify that artificial intelligence can err. Students should therefore verify the results returned to them with the help of others and are responsible for any errors or omissions that the tool provides. Additionally, students must show intellectual honesty by citing their source (which happens to be ChatGPT) as one does when compiling a bibliography. Failure to do so constitutes a violation of the principles of academic honesty.
Arvind Narayanan from Princeton University believes that ChatGPT is nothing revolutionary:
Sayash Kapoor and I call him a bullshit generator, as others have. We don’t mean that in a normative sense, but in a relatively precise sense. We think that he is trained to produce plausible texts. He’s very good at persuasion, but he’s not trained to make true statements. He often produces true claims as a side effect of his plausibility and persuasiveness, but that is not his goal.
It is in fact what the philosopher Harry Frankfurt called bullshit, a speech intended to convince without worrying about the truth. A human storyteller doesn’t care if what he says is true or not; He has specific goals in mind. As long as he is convincing, these goals will be achieved. This is exactly what ChatGPT does. He’s trying to be persuasive, and he has no way of knowing for sure whether the statements he’s making are true or not.
What about Bard’s Microsoft-funded OpenAI Google ChatGPT response?
Earlier last week, Google announced its AI Bard chatbot. But the bot didn’t get off to a good start, as experts noted that Bard made a factual error in his very first demo.
A GIF shared by Google shows Bard answering the question: What new discoveries from the James Webb Space Telescope can I tell my 9-year-old about? Bard offers a three-point enumeration, including one point indicating that the telescope captured the first-ever images of a planet outside our own solar system.
However, a number of astronomers pointed out on Twitter that this was wrong and that the first image of an exoplant was taken in 2004 – as stated on the NASA website (see source): This is not meant to be an asshole (actually , if ultimately ) and I’m sure Bard will be impressive, but for the record, JWST hasn’t captured “the first-ever image of a planet outside our solar system,” tweeted astrophysicist Grant Tremblay said.
Chatbots: Fairy Dust?
Recent developments are raising questions that make us decide to use the search engine, whether Google or Microsoft: should we really rely on a chatbot to search the web for information? Is the approach better than the old one, which consists of letting humans go in search of information in order to synthesize it?
In fact, chatbots like ChatGPT have a well-documented tendency to present false information as fact. Researchers have been warning of this problem for years. This is also why some teachers have instituted open guidelines for using ChatGPT, telling their students that artificial intelligence can be wrong. Students should therefore verify the results returned to them with the help of others and are responsible for any errors or omissions that the tool provides.
And you ?
What is your opinion on the topic?
See also:
AI researchers say OpenAI’s GPT-4 language model could pass the bar exam, sparking debate about replacing lawyers and judges with AI systems
Microsoft is investing $1 billion in OpenAI, the company founded by Elon Musk that is trying to develop AI similar to human intelligence
Google trained a language model that could answer medical questions with 92.6% accuracy, the doctors themselves scored 92.9%.
Microsoft creates code autocompleters using GPT-3, OpenAI’s text generation system, to fill the developer shortage worldwide