The world is humming, perhaps even confused, about the increasing use of artificial intelligence. One of the most popular artificial intelligence (AI) tools available to the public today is ChatGPT, an AI-based language model that is “trained” and provides vast amounts of online information. After getting all this, ChatGPT can reproduce human text responses for a given prompt. Can answer questions, discuss many topics, and write texts.

It’s not hard to imagine a robot navigating the surface of Mars, connected at the factory to ChatGPT or a similar artificial intelligence language model. This smart bot can be loaded with a range of scientific instruments. He could analyze what his scientific tools found “in situ”, perhaps even matching any past-life evidence he discovered almost instantly.

These data can be digested, evaluated, evaluated and compiled in some scientific form. In a well-organized condition with additional footnotes, the product can be submitted for publication directly from the robot to a scientific journal such as Science or Nature. Of course, this article will be peer reviewed later – possibly by AI/ChatGPT reviewers. I contacted several leading researchers who presented this extraterrestrial and Martian scenario with mixed reactions.

prone to hallucinations

Serkan Özcan, a lecturer in the Department of Innovation and Technology Management at the University of Portsmouth in England, said, “It can be done, but the information can be misleading.” “ChatGPT is not 100% accurate and prone to ‘hallucinations’.”

Ozcan said he wasn’t sure whether it would be useful without a prior working group to analyze and emulate ChatGPT. “I think people can do better than ChatGPT even though it’s slower,” he said. His advice is not to use ChatGPT in “areas where we can’t accept errors”.

people are aware

Steve Ruff, associate professor in the School of Earth and Space Studies at Arizona State University in Tempe, Arizona, is deeply involved in Mars exploration.

“Given that the process involved discussion among the team about the observations and their interpretation, my initial reaction was that manuscripts ‘in situ’ are very unlikely to be a realistic scenario,” said Ruff. “I doubt that any AI trained on existing observations can be used to confidently interpret new observations without human intervention, especially with new datasets that were not previously available. A painstaking effort is required to understand each such dataset.”

In the near future, Ruff believes that artificial intelligence could be used for the work of Mars rovers, for example, to select targets for observation and navigation without the participation of humans.

Firstly

What kind of world do we want to live in? This is perhaps the most pressing question, said Natalie Cabrol, director of the Carl Sagan Research Center at the SETI Institute in Mountain View, California.

“About the real issue,” Cabrol said. “Artificial intelligence is a powerful tool and should be used as such to support people in their activities. In fact, we do this every day in one way or another,” he added, “and improved versions can make things better.”

On the other hand, like all human tools, they’re a double-edged sword, and sometimes people start to think they’re “nonsense,” Cabrol added, and he thinks that’s the case here.

“Personally, I love writing articles. It’s a great time to see my work come to life and get my ideas down on paper,” Cabrol says, and sees it as an important part of his creative process.

“But let’s suppose for a moment that I let this algorithm write it down for me. If you let the algorithms do the work for you, it’s because you assume they’ll be less biased and do a better job. With that logic, I would assume that person does not have the authority to review this article.”

Ghosts of “transhumanism”

Cabrol senses the question: where do we stand? What if all researchers ask AI to write research grant proposals? What if they do and don’t tell?

“It depends on what kind of world you want to live in and what part you want to leave to humanity,” Cabrol said. “We’re creative creatures and we’re not perfect,” he continued, “but we learn from our mistakes, and that’s part of our evolution. Error and learning are just other words for “adaptation,” he said.

We interfere with our own evolution by allowing artificial intelligence to penetrate what makes us human, Cabrol added, and he sees the specter of “transhumanism” in all this. Transhumanism can be defined as a free ideological movement combined with the belief that the human race can go beyond its current physical and mental limitations, especially with the help of science and technology.

“Of course you will say it’s not a chip in our brain, it’s just paper. Unfortunately, this is part of a much broader and very disturbing discourse about the (mis)use of artificial intelligence,” Cabrol summed up. It’s not just paper. It’s about what we really want to be as a species. Personally, I see AI as a useful tool and will limit myself to that.”

knowledge gap

“It’s funny that we’re still arguing about the definition of life as we know it, and in that quest we’re starting to use a tool that expands the definition of life,” said Amy Williams, an associate professor at the university. Department of Geological Sciences. Florida in Gainesville. He is a scientist-participant in NASA’s Curiosity and Perseverance rover, where robots explore Mars.

Williams reacted to AI-ChatGPT’s extraterrestrial environment in full disclosure mode. “The first time I used ChatGPT, in preparation for this answer, I asked him: “What organic molecules did the rovers find?” The question was based on my particular area of ​​expertise,” he told Space.com.

“It was illuminating because it did a great job of providing what I would call reliable and appropriate statements for a resume I could have in an educational conversation with the general public about organic molecules on Mars,” Williams said.

But he also showed Williams his limitations as he could only access data from September 2021 – labeling it an “information outage”.

“So their answer doesn’t cover all published results about organic matter on Mars that I’ve known since 2021,” he said.

Emphasizing that he is not an expert in artificial intelligence or machine learning, Williams said that future iterations of ChatGPT + AI could likely incorporate newer data and create a complete synthesis of recent findings from any scientific study.

“However, I still see them as tools that can be used. with not by people in its place “Given the data limitations on the uplinks and downlinks of our current deep space network, I’m trying to find a way to load a knowledge base for something complex, such as current and historical data and sources, recipients and content. “The fate of organic molecules in Marcy so that the on-board artificial intelligence can create a manuscript for publication.”