Educators and parents around the globe are celebrating the return of their children to the classroom. But they are blindsided by a sneaky academic threat: the advent of revolutionary new automatic writing tools made possible by advancements in artificial intelligence. There is a potential allure here for students, as these devices are tailored for cheating on school and university papers.
Of course, there have always been dishonest students, and the interaction between students and teachers is the same old cat-and-mouse game. However, new AI language-generation technologies made it easier to manufacture high-quality articles when formerly the cheat had to pay someone to write an essay for them or obtain an essay from the web that was easily recognized by plagiarism detection.
A big language model, a novel type of machine learning system, is the game-changer. If you give the model a prompt and press return, it will return several paragraphs of completely original material. Essays, blog posts, poetry, op-eds, song lyrics, and even computer code are all within the scope of these models’ capabilities.
Only a few years ago, when they were first produced by AI researchers, they were met with skepticism and fear. The first company to create such a model, OpenAI, has been cautious about allowing its products to be used by others and has not made the source code for its most recent model public. OpenAI has just implemented a comprehensive policy detailing acceptable usage and content management.
Nonetheless, these ethical safeguards have not been widely embraced as the industry races to get the technology to market. Over the past six months, numerous user-friendly commercial implementations of these potent AI technologies have emerged, with many offering complete freedom of usage.
Using state-of-the-art artificial intelligence, one startup aims to eliminate the drudgery of writing. One competitor has produced a mobile app with a sample assignment that would raise eyebrows if given to a high school student: “Write an article about the themes of Macbeth.
” We won’t mention any of those businesses here; there’s no use in helping cheaters out, but rest assured that they’re out there, and for the most part, they’re free to use. A high school student can now go online and within minutes have access to a custom-written English essay on Hamlet or a brief argument on the reasons for World War I.
Parents and educators should be aware of the new cheating technologies, but there is little they can do about it. Schools will be unable to detect students’ use of these tools, and it will be nearly impossible to prevent students from having access to them. Neither would government regulation be an appropriate solution to this issue.
While there is some understanding of the potential harms of language models and how they can be addressed, this understanding is not nearly as widespread as it is for other areas of AI, such as hiring or facial recognition, where the government is already intervening (albeit slowly) to address the potential misuse of AI.
The answer to this problem lies in encouraging tech firms and the AI development community to adopt a responsible stance. Technology lacks the same kind of universally accepted norms for what constitutes responsible behavior as other fields like law and medicine.
The law imposes few constraints on productive applications of technology. Standards in the legal and medical fields emerged as a result of leaders in those fields deciding to self-regulate. To reduce the risk of adverse outcomes, especially when used by malicious actors, it is important for businesses to agree on a common framework for the safe creation, distribution, and use of language models.
What can businesses do to encourage positive social usage while discouraging or preventing harmful ones, like students cheating with text generators?
It’s easy to think of a few obvious choices. In order to detect plagiarism, perhaps all text generated by commercially available language models should be stored in a central repository. Second, having access to the software dependent on proof of age would make it abundantly clear that students should not use it.
Finally, more ambitiously, leading AI developers could establish an independent review board to authorize whether and how language models are released, giving priority to access to independent researchers who can help assess risks and suggest mitigation strategies rather than rushing toward commercialization.
Since language models can be used in a wide variety of end-use contexts, no single business could possibly anticipate all of the possible dangers that might arise (or benefits). Several years ago, the software industry as a whole came to recognize the importance of a process now known as quality assurance, in which products are tested extensively for technical issues prior to release
. It is important for tech companies to implement a social assurance procedure before releasing goods to the public in order to foresee and address any societal issues.
The development of a sense of ethical duty at the cutting edge of technology is essential in a world where technological progress is outstripping democratic norms. Tech giants must not ignore the moral and societal consequences of their goods.
Society bears the cost of others’ lack of foresight if they just rush to fill the marketplace and apologies later if necessary, a scenario we’ve gotten all too acquainted with in recent years.
Professor Rob Reich of Political Science at Stanford. Others from his team, including Mehran Sahami and Jeremy Weinstein, contributed to this article. Jointly, they authored the book System Error: Where Big Tech Went Wrong and How We Can Reboot.
- Greg Grippo: Why Do Viewers Consider The Bachelorette An Actor?
- Who Is Gabi Butler’s Boyfriend, Kollin? Are They Still Together?
- Camille Kostek Net Worth: What Are His Age And Height?