No, GPT-4o isn’t coming for your job. Yet.

It’s easy to see why it’s creating such a stir: As demonstrated on Monday, the assistant can detect sarcasm and speak with emotion, creating an intelligent persona that’s being compared to the advanced Scarlett Johansson-voiced AI of the 2013 film “Her.”

But while this version of ChatGPT might appear far more accessible and sophisticated than its competitors, it’s unlikely to eliminate professions soon, AI experts told Business Insider.

They also cautioned that most people have been reacting to a tech demo, which can often be highly tailored and might not necessarily reflect the product’s true capability.

Even if OpenAI’s new ChatGPT is as powerful as advertised, it’s more likely roles in these fields will be tweaked while the professions stay intact, they told BI.

What’s getting people so riled up about GPT-4o?

Clips from Monday’s demo make ChatGPT look like a massive leap for artificial intelligence and how we can use it in the real world.

One X user said he might start using it as a personal assistant to make phone calls on his behalf.

OpenAI also showed potential adopters new use cases, such as translation and teaching services.

The company posted a demo video showing a group of OpenAI employees, including chief technology officer Mira Murati, asking the assistant to help with a math problem.

The assistant didn’t solve the algebra equation. Instead, it walked presenters through each step of the problem, giving real-time instructions like: “Get all the terms with X on one side and the constants on the other side.”

All the while, it responded to feedback and follow-ups whenever the presenters spoke up.

Another clip of OpenAI’s demo was posted by a tech journalist who wrote: “RIP translators.”

In the clip, ChatGPT appeared to seamlessly translate between Murati, who spoke in Italian, and an OpenAI employee speaking in English.

“When Siri and Alexa were first launched, they did cause concern about job security in similar areas,” said Daan van Rossum, the founder of FlexOS, a Singapore-based, work-focused media company. “None of those AIs were even close to human-like. ChatGPT4-o changes all of that. The difference can’t be understated.”

The new video camera and voice features do wonders for ChatGPT’s potential in roles like customer support, van Rossum added.

Companies and industries are already preparing for that AI disruption. In January, for example, language learning app Duolingo axed 10% of its contract roles, saying that generative AI will be used to create more content moving forward.

Why these jobs aren’t going to disappear

It’s unlikely that ChatGPT’s new personal touch will eliminate professions in teaching, translation, and customer support, said Leslie Teo, senior director of AI products at AI Singapore, the country’s national AI program.

That’s precisely because these jobs need a personal touch, he said.

“There’s something about human empathy,” Teo said, highlighting that teachers have gone through the learning process themselves and understand how humans struggle, unlike AI. “Those are very powerful things.”

Teo said that people working in teaching, translation, or customer support are more likely to use AI to make their roles easier, rather than be supplanted by technology.

“A good customer service agent exudes empathy and understanding, something that a machine cannot do,” he added.

Ben Leong, a computer science professor at the National University of Singapore, said it’s too early to draw conclusions from just a demo. He said he foresees interpreters’ businesses being majorly disrupted.

“AI is really good for well-defined problems. Translation is a very well-defined problem,” he said.

But it’s going to be much more difficult for ChatGPT to take over customer support, teaching, or negotiation, Leong said.

“If someone negotiates a bad deal for you, who is responsible? Are you liable? Can AI be your legal representative?” he said.

Simon Lucey, the director of the University of Adelaide’s Australian Institute for Machine Learning, said ChatGPT still makes errors when multiplying numbers of three digits or more, even in its later versions.

“So at the moment, it’s a helper. At the end of the day, if you want to get something meaningful or do something reliably, you still need to get a human to cast desire over what it’s producing,” Lucey said.

Some people may lose their jobs in the next few years, Lucey said, but he added that they have reason to stay optimistic.

“In the 1980s, when Microsoft released Excel, people were petrified and said it would put all of these accountants out of a job. We’ve got more accountants now than in the 1980s,” he said.

Teo agreed. “Many roles will be replaced by AI, but that doesn’t mean jobs. Our jobs are bundles of roles,” he said.