The new possibilities of smarter ChatGPT also raise concerns

The advanced text generator ChatGPT caused a stir in November last year. It suddenly became clear how good AI (artificial intelligence) is already. To the surprise of the creators themselves, who brought the chatbot online among the planned works. This planned work is an improved version released this week: GPT-4.

It has long been clear that artificial intelligence is one of the most important developments in the technology industry. Since last month, the strong position of Google’s search engine has suddenly become a topic of discussion due to competition from Microsoft.

ChatGPT has integrated it with the Bing search engine. Neither of these two tech companies is standing still and released updates this week. It shows that developments in this field are progressing rapidly and that the potential is great.

Differences between 3,5 and 4

There are some fundamental differences between GPT-4 and the previous 3.5 version. First, the new version can handle a lot more words, allowing the system to “read” a lot more and produce a lot more text.

NOS asked three questions each to GPT-3.5 and GPT-4. The answers were (due to length limitations, the question is the maximum number of words):

The most striking innovation is not yet available, but it is delicate. GPT-4 can recognize what’s in an image and then do something with it. The general public cannot yet test this for fear of harassment, but the potential has already been clarified.

OpenAI President Greg Brockman showed a sketch of a website from a notebook during a presentation that lasted more than 20 minutes this week. He then took a picture of it and gave it to GPT-4 with the task: Create a website. That’s what happened, including working buttons.

“It seems like magic to the outside world: You go from the sketch to the construction site,” answers AI researcher Judith van Stegeren. As a PhD student, she spent four years researching creative artificial intelligence at the University of Twente.

“Secretly, it converts from photo to text, and that text is then linked to a task to code a website. They take one small step after another.” He also states that a lot of research has been done on image recognition and it has been successful in the early period.

From the sketch to the website:

ChatGPT started a search engine war between Google and Microsoft this year. For the first time, market dominance is discussed. But that’s not all: the parties are also on equal footing when it comes to Office apps. Gmail and Outlook, Word and Google Docs.

And just like in February, companies once again fell for each other to show users what they had to offer. You may then want to summarize the email conversations or turn them into a briefing or presentation. Or how about a summary of a video conference you attended later?

dark side

In addition to the potential, this AI development also has a downside. A few may even be said. For example, the text generator can be abused to spread fake news.

And how do you deal with a system that will make mistakes? Perhaps on important company information. Human control remains essential. As a user, you should always be careful that the system does not write anything wrong. It is not for nothing that OpenAI emphasizes that humans and computers should do this together.

Tech ethicist Olya Kudina, an assistant professor at TU Delft, is also concerned about the speed with which developments are currently taking place. “You still need to learn how to use this technology. This creates an unfair urgency to integrate them into your own products now, otherwise it will be too late.”

You think this is a great socio-technical experiment. “It’s like OpenAI is using the community as a big lab where people can see and consider their reactions.”

However, there is no talk of the withdrawal of natural gas at the moment. Alongside Microsoft and Google, communication platforms Slack and Discord, payment service Stripe and language app Duolingo are also working on ChatGPT integration.

Source: NOS