Sam Altman, CEO of OpenAI, the company behind ChatGPT responds to popular concerns about the tool. There are valid concerns that the AI chat robot, ChatGPT may be used to cheat in exams. However, he claims that OpenAI will not fold its arms while things go south. In a recent interview, Altman claims that OpenAI will develop methods to help schools identify AI cheating. But he cautioned that there was no guarantee cheating would be fully detected. He said
“We’re going to try to do something in the short term, maybe some way to make it easier for schools to detect cheating in systems like ChatGPT. But honestly, some people may find a way to get around the detection,”
He adds: “For a long time, people have been incorporating new techs into their lives as well as into the classroom. These techs can only have a more positive impact on users. Generative text, for example, is something we all need to adapt to. I think, we adapted the calculator and changed what we tested in math class. ChatGPT is undoubtedly the more extreme tech, but the benefits it brings are also greater”.
New York bans students from using ChatGPT
Prior to Altman’s remarks, the New York City Department of Education and the Seattle Public Schools system had banned students and teachers from using ChatGPT to prevent plagiarism and cheating. These bans sparked discussions about how AI could change the state of education and the way students learn, especially among teachers. “I can see why educators feel this way, and we’re likely to see the same in many other fields before long,” Altman said.
Altman said that although OpenAI has heard from some teachers about the impact of ChatGPT on students’ homework and other aspects, many teachers believe that chatbots can become “every child’s most competent personal tutor. “
In fact, Altman thinks using ChatGPT can be a more engaging way to learn. “I use it myself to learn new things and find it more engaging than other ways of learning. I’d rather let ChatGPT teach me something than read a textbook,” he said.
Altman claims that OpenAI will experiment with watermarking and other techniques to label content generated by ChatGPT. However, he warns schools and national policymakers to avoid relying on such tools. “Fundamentally, I don’t think it can be perfect,” Altman said. “People need to figure out how much text they need to modify, and of course, there are other things that modify the output text.”
Given the growing popularity of ChatGPT, Altman believes the world must adapt to the presence of AI. He believes that with time, the tech will improve so that it can work only where it needs to. “It’s a changing world and we all have to learn to adapt. I think it’s better that way. We don’t want to go back to the past,” he said.
Gizchina News of the week
GPT-3 outperforms human college students in IQ tests
GPT-3 is a generative AI system that helps with text generation. After training about 200 billion words and burning tens of millions of dollars, the most powerful AI model in history “Generative Trained Transformation Model 3” (GPT-3) was born.
The language AI model created by OpenAI is like an all-rounder. It only has painting styles that you can’t imagine, and there is no copy that it can’t output. It can create literature, translate, and write its own computer code. Any layman can use this model, provide examples, and get the desired text output in a matter of minutes.
According to reports and reviews of this system, some members of staff at the University of California, Los Angeles found that in a series of tests that measure intelligence (IQ), the autoregressive language model of GPT-3 did much better than a regular college student. This is why teachers worry that this system will take over assignments for students.
The program uses deep learning to generate text that resembles human language. GPT-3 has many uses, including language translation and text generation for chatbots. Also, with 175 billion parameters, it is one of the largest and most powerful language-processing AI models available.
ChatGPT has a similar effect
It suddenly occurred to me here that OpenAI’s ChatGPT seems to have achieved a similar effect. Although it is still based on GPT-3, this model is “GPT-4” in the industry. This is also the fourth generation of the Silicon Valley Research Laboratory. Language models pose an existential threat to search engines, writers, coders, professors, and Nickelback’s all over the world. Of course, in the opinion of most experts, the current version of ChatGPT is only an appetizer compared to the upcoming major version of GPT-4.
Researchers at the University of California believe that such large language models have reignited the debate about whether human cognition is stronger when provided with enough training data. Of particular interest is the ability of these models to reason about new problems at zero-shot, without any direct training on these problems.
In human cognition, this ability is closely related to analogical reasoning, the researchers noted, while they conducted direct comparisons to GPT-3 on a range of analogy tasks, including a novel text-based approach closely related to Raven’s Progressive Matrix. They finally found that GPT-3 showed amazing abstract pattern induction ability, matching or even surpassing human ability in most cases. The final results show that large language models such as GPT-3 have acquired an “emerging ability” to find zero-shot solutions to a wide range of analogy problems.