The dangers of AI as told by OpenAI’s CEO
It isn’t an exaggeration to say that ChatGPT has taken the world by storm. It took just 5 days for the Open AI chatbot to reach one million users, which is incredibly fast. In comparison, it Netflix three and a half years, Facebook ten months, Spotify five months, and Instagram two and half months to reach the same number of users that ChatGPT got to in less than a month. We have previously examined all the ChatGPT hype and the toll itself, but OpenAI’s CEO Sam Altman recently took to Twitter himself, with a warning about his company’s own product.
Altman said in his tweet:
“ChatGPT is incredibly limited, but good enough at some things to create a misleading impression of greatness.
it's a mistake to be relying on it for anything important right now. it’s a preview of progress; we have lots of work to do on robustness and truthfulness.”
Interestingly, Altman’s comments echo our own conclusion when we looked at the new tool over a week ago. We said:
“The value of ChatGPT very much comes from how you are using it and what you are using it for. It is an incredibly useful and even entertaining tool in some instances, but you shouldn’t be leaning on it too heavily as it will likely creak and snap when put under any real pressure.”
The confirmation from Altman then, only backs up what we have already said. The tool is fun to use and is able to offer inspiration but it should not be relied upon for factual accuracy and is not a dependable ally when researching or creating outbound content. In short, the tool is a quick and reliable assistant that can point you in the right direction, but you must always take those final steps yourself.
This is a common theme with the AI-based tools we are seeing today. Two other examples, that spring to mind here are Jasper and Lex that both offer AI copywriting services. Jasper comes via a pretty hefty subscription model while Lex is free and has also been optimized for use as a Google Docs alternative on mobile. Both allow you to input various commands and prompts in a similar fashion to what you can do on ChatGPT and will then output various levels of content, automatically created based on your prompts. These are incredibly useful writing tools and can be good for quickly structuring articles and filling in sections and even paragraphs. However, the second you send out one of these AI-created pieces of writing without checking, whether that be to publish online or send as an email you are taking your credibility for granted.
That brings us to the important lesson to remember then with these tools. They are not writers, and no matter the hype, ChatGPT is no more than a language model that is leaning on training data. The key word to remember here is tools. They are new tools that offer us new ways of doing things. They are not something we can rely on to do jobs for us. These tools offer a lot of promise for the future and utility for the present, but it is important to know what they are so that you can know how to use them properly.
They would endanger us if “the elites” let them.
Far more artificial than intelligent. What is intelligence by the way? To understand a question, to bring an answer? What answer? If the answer is to state “I understand your question but cannot at this time answer it [bla-bla]” then any 2-line script can do that with the difference of a data bank’s volume. Rather elaborated automatisation as i see it.
I don’t deny that so-called AI is a work in progress and will increasingly participate to complex tasks but considering it as intelligent is more, as I perceive it, an elaborated automatisation network than a haven of deep thoughts. When I read that it could be that one day trials would be processed with AI I remain stunned, immaculate.
Literature, poetry, philosophy, music (apart from dance music which sounds totally artificial)… I cannot perceive there a place for AI. Mathematics, physics probably yes, though depends of the approach. Deductive certainly but inductive intelligence (an idea born from nothing besides one’s brain’s own language as Lakan stated it, on which you then work to bring validity evidence) leaves me skeptical.
Jacques LACAN, not LAKAN of course :=)
I would disagree with your differering classifications of disciplines. Music or Math, Philosophy or Physics, all take the same ‘tools’ only a few have to be expressed or understood in ways no one has thought of yet.
@Tachy, a true debate! I cannot agree that “Music or Math, Philosophy or Physics, all take the same ‘tools’ only a few have to be expressed or understood in ways no one has thought of yet.”
“Tools” to start with. Picasso wrote “I don’t search, I find”. Perhaps was he referring to the fact that he considered that searching involved an artist’s quest on the basis of “tools” as you write it, but that imagination, creativity (let’s get the word loose) precisely was not the result of an equation (but that perhaps of “inductive” intelligence as I mentioned it above). Intelligence as i perceive it is far more than what IQ may consider it to be.
Take for instance intuition : is it an element of intelligence? French philosopher Claude Tresmontant wrote in one of his books that he considered intuition to be “a converted intelligence capable of thinking the real in the sense of its genesis”. If so : AI, beware!
If I don’t know what intelligence is — and I definitely don’t — then neither do I know what it is not. Let’s put it this way : my comment is only my two cents perception, perhaps on the basis that intelligence seems to me such a tremendous area that exposing it to tools appears to me as a shrinking factory approach. Seems.
Playing with all these new OpenAI AI projects has been a lot of fun. There will be no need for artists anymore if people keep training their databases. Twitter is already filled with sobs from artists like Karla Ortiz.