GPT-4’s launch ‘another step change’ for AI and higher education

New version of OpenAI’s chatbot can interpret images and tutor students

March 23, 2023
Final touches to the winter 2019 exhibition at Somerset House, “24/7: A Wake-Up Call For Our Non-Stop World”
Source: Alamy

The ability of the new version of ChatGPT to interpret images and “tutor” students poses new challenges and opportunities for higher education, according to experts.

With universities still scrambling to understand the impact artificial intelligence (AI) chatbots might have on assessments and research, OpenAI has released GPT-4, just 15 weeks after its previous version caused such a stir around the world.

Although full access to the technology was still restricted, it appeared to represent another “step change” in what generative AI was capable of, according to Thomas Lancaster, a senior teaching fellow in computing at Imperial College London.

In its publicity around the launch, OpenAI highlighted GPT-4’s increased ability to pass popular examinations compared with ChatGPT, claiming it scored 298 out of 400 in the bar exam and 163 in the Law School Admissions Test, compared with 213 and 149 with the previous version. The risk of students using AI to write essays or complete exams has already led to many universities banning ChatGPT on campuses.

The reported results were “incredible”, said Dr Lancaster, but it was the ability to process and respond to images that set GPT-4 apart from what came before, he said.

“That means you can take a photo of an exam question with pictures and diagrams and expect GPT-4 to be able to formulate a response.

“That might not be useful in an invigilated exam where students don’t have easy access to GPT-4, but there are all kinds of real-world uses of this technology that we need to be preparing students for.”

Mike Sharples, emeritus professor at the Open University’s Institute of Educational Technology, agreed that the image-interpretation facility presented several potential new uses of generative AI.

“Students will be able to use it to answer complex exam questions containing figures and illustrations. They will also be able to use GPT to summarise entire research papers,” he said.

“It could interpret concept maps, flowcharts, even cuneiform text. It can turn a sketch of a computer interface into a working computer programme.”

Professor Sharples said GPT-4 could also act as a “knowledgeable guide” to students because it could respond to queries in an intellectual way, guiding and shaping their learning.

Education company Khan Academy has already announced that it will use the new technology to power a virtual tutor, and Professor Sharples said this “transformation shouldn’t be underestimated”.

Academics could also use this new feature in their work, he added, and he himself was using GPT-4 to “augment my thinking in preparing talks and papers”.

The fact that GPT-4 can now generate up to 25,000 words in one go was also a “huge opportunity and a huge challenge”, according to Professor Sharples.

“It means that academics, and students, can ask GPT-4 to generate works up to 50 pages in length,” he said, adding that this was a “big challenge to the entire academic publishing process”.

He said he had used GPT-4 to generate an entire “conference paper” on comparisons between college student essays and AI-written versions, giving it only a title and a list of main research questions to address as prompts.

It successfully “concocted a research study, with methods and results, including a table of findings and a statistical analysis of the invented data”, Professor Sharples said, adding that this should force journal editors and conference chairs to rethink their acceptance criteria for papers.

For example, they might now need to require access to the full data gathered by the researcher, he said, which would bring potential privacy issues, and develop new guidelines to clearly state what is and isn’t acceptable with regard to how AI is used by the author.

Dr Lancaster said the writing ability of GPT-4 also appeared to have advanced, addressing one of the issues identified with ChatGPT, which tended to produce formulaic sentences that could then be picked up by AI detectors developed to identify AI-generated text.

David Cyrus, one of the creators of an AI detector that has emerged recently, AICheatCheck, said the tool had been trained to prepare for GPT-4 and it appeared to work just as well, based on a test of 400 articles.

But Dr Lancaster said GPT-4 could “write much more in the style of an individual author, as well as being able to write longform”, meaning it would be harder for detection systems trained on ChatGPT to function. 

“It also means that academics comparing student writing styles between assessments might no longer use that approach to detect academic misconduct,” he added.

“I’d expect to see continual advances in this technology – so, more than ever, we need to work with this in the future, rather than against it. This is an incredibly exciting time to be working in education.”

tom.williams@timeshighereducation.com

POSTSCRIPT:

Print headline: ChatGPT-4 brings step change, but also a challenge for universities

Register to continue

Why register?

  • Registration is free and only takes a moment
  • Once registered, you can read 3 articles a month
  • Sign up for our newsletter
Register
Please Login or Register to read this article.

Related articles

The AI chatbot may soon kill the undergraduate essay, but its transformation of research could be equally seismic. Jack Grove examines how ChatGPT is already disrupting scholarly practices and where the technology may eventually take researchers – for good or ill

16 March

Reader's comments (4)

"More than ever, we need to work with this in the future, rather than against it. This is an incredibly exciting time to be working in education", said Dr Lancaster from his local job centre whilst applying for jobs stacking shelves in Tesco....
I’m so relieved to have recently retired. The issue is not whether these AI systems can be used to quickly and effectively generate text that provides ideas for the structure or scope of an augment; they clearly can. The issue that would worry me as an educator is how to motivate students to develop their own skills in this domain. Staff who have the skills over many years have been quick to point out the weakness of these tools but few seem to have considered their impact on students who are faced with the prospect of many iterations of trial and error as they develop their own voice within the confines of discipline-specific writing styles. I’d be happy for anyone to use these tools but if my role within education includes warranting students’ own capacity to structure written arguments and to blend into their writing well chosen relevant evidence, then the writing I have to see must be the students’ own attempt and not text that has been initiated or structured by AI tools. Only those academics who have the luxury of small groups of students in regular tutorial will be able to determine by discussion whether a student has developed their own written work. In UK HE, very few academics have such regular meetings with the same students at a frequency high enough to develop the necessary familiarly.
I started teaching college students in business and politics faculties 10 years ago, after a career in law. In recent years my exams have had a higher proportion of essay questions than when I started out. So my immediate reaction to GPT-3, -3.5 and especially -4 was overwhelmingly negative (other than for the limited purpose of asking students to generate a bot-written essay, and then critique it). Reading articles like this gives me pause, making me wonder if I'm on the wrong side of history. After all, when I was in (US) law school 40 years ago, profs inveighed against using commercial outlines to prepare for exams in the standard subjects (property, torts, etc.). For my first (1L) year I believed them, analysed everything on my own, and did OK. As a 2L, I used outlines, and got straight As. It was so much more efficient than analyzing cases on my own (though I had learned that skill in my 1st year). Maybe I'm being like my old profs? No. First of all, the information in the better commercial outlines was correct -- one can't rely on even GPT-4 to that extent. Second, what the outlines did was much more limited: give you the summary facts and key holding of a case, and the more general rule derived from it. A student still had to figure out what were the issues in a novel fact pattern, which bodies of law were relevant, which precedents and rules might be relevant, what analogies might be needed, how to organize all that disparate information into clear and cogent arguments for each side, and what outcomes were the most likely. GPT-4 eliminates the need for all of those skills, from the get-go, in law and all other fields -- and potentially mixes in false information, to the bargain. A student who starts relying on the bot sooner will never develop the skills needed to critique its output. Now comes the AI's ability to mimic individual writing and rhetorical styles, coupled with being hard to detect. Only people who deliberately insulate themselves from domestic and world news could sincerely believe that the benefits of this outweigh the potential harms. Which leads one to conclude that tech companies already know that the real money to be made from bots is in weaponizing them on the dime of the forces of darkness. More forbearance in rolling out these bots might have allowed some advance public discussion of how to integrate them into society. Instead, the market pressures of capitalism push the bots' corporate owners to dump this stuff steaming into our laps every few weeks. How can those of us who aren't despots possibly adjust to that? And why ought we have to adjust to it? This is plainly a reckless path for our society. The cheerleaders about "exciting times" do us no service. There were cheerleaders for the dotcoms and for credit derivatives earlier in this century, and those trends came crashing down on us all. When that happens with these bots, there will again be a couple of years of op-eds and panel discussions along the lines of "Who could have known?" It's obvious today: this is a bad idea.
I am an English teacher and I got tired of finding tools to prove that an essay or any other text was not copied, but written by the student, so if you can't beat them, join them. I am using chatGPT to lead in different ways learning activities in class and especially working with prompts that allow the student to compare text, analyze points of view, error correction and in that way keep working the reflexive thinking, but most importantly, teach them the appropriate use of this powerful tool. Honesty comes first.

Sponsored