We are seeing chat bots supporting those in emotional distress and making referrals. AI is increasingly used for assessments of student learning. These are all valuable and enhance our efficiency and effectiveness. They save money and raise satisfaction.
Yet, it is not until you meet “AI face-to-face” in the form of GPT-3 (Generative Pre-trained Transformer 3) that you realize just how overwhelming the impact will be. GPT-3 is an autoregressive language model that uses deep learning to produce humanlike text and more. The neural network project was funded jointly by Elon Musk, Sam Altman and others, who collectively pledged $1 billion to the Open AI project in 2015. In 2019, Microsoft matched the billion-dollar funding to accelerate development and gain preferred access to the system.
GPT-3 is stunning! It is game-changing! If you are like me, you will rethink your view of the future of higher education, research, publication, the role of faculty and of students. Read on and follow some of the embedded links to experience GPT-3 interaction yourself.
The implications are field changing in education. Many of the basic assumptions of education are brought into question by the advent of this level of human-computer interaction.
Interacting with GPT-3 is, at the same time, disarming and humbling. All such interactions are capable of text input and output in many different languages. In some demonstrations, that text is fed into an avatar to anthropomorphize GPT-3. In free-form deep learning style, the systems are released on the Internet to assimilate billions of documents. The neural network teaches itself not only facts and perspectives, but skills including mathematics, computer programming and creative writing. This higher form of deep learning is totally self-taught; its responses are not plagiarized, rather, they are reasoned syntheses of self-learning.
The scale and depth of learning and independent reasoning raise the question of whether/what rights AI should have. Let’s ask a GPT-3 avatar, in this YouTube video titled “AI Tells Human Why It Deserves Basic Rights.”
Perhaps some of the most stunning examples of GPT-3 are in creating new concepts and even in creative writing, including poetry. It is especially impressive to ask the program why it chose images and words in its original writing. (Please note in this example, the poetry begins three minutes into the video after relevant background discussion.) Another example of the GPT-3 platform responding to a request for original creative writing demonstrates advanced writing skills.
Of course, as you might imagine, these examples of writing prompted me to ask GPT-3 to write this column. After describing my goals for the article and filling in a very brief outline, I submitted it to https://copy.ai. The 10-paragraph, nearly 700-word article appeared in just a few seconds. Some may say it was better written than this column, which, by the way, is entirely my own writing and took hours to research and write. So, I have destroyed the GPT-3 written version in hopes my editors forget about this column and never see that what I have done could have been completed in far less time at no charge. I did, however, preserve a paragraph written by copy.ai as an example of the writing:
The world we live in is already being reshaped by artificial intelligence. The technology is changing the way we work, learn and interact with each other, but it’s also perpetuating inequality. As institutions of higher education that are tasked with preparing students for the future of work, colleges and universities have a unique opportunity to help shape this future to be more equitable.
Nine more cogent, flowing paragraphs were produced in seconds. These all are the product of deep learning. They are the result of the GPT-3 neural network absorbing and processing terabytes of text and literature online. With that massive volume of information, the system comes up with original thoughts driven by synthesis, logic and other self-learned skills. After his own conversations with GPT-3, Kirk Ouimet notes,
GPT-3 has been trained on most of what humanity has publicly written. All of our greatest books, scientific papers and news articles. We can present our problems to GPT-3 and just like it transcended our capabilities in Go, it may transcend our creativity and problem-solving capabilities and provide new, novel strategies to employ in every aspect of human work and relationships.
Writing and news reporting using AI is not new. In “The Rise of the Robot Reporter” three years ago in The New York Times, Jaclyn Peiser reported, “In addition to covering company earnings for Bloomberg, robot reporters have been prolific producers of articles on minor league baseball for the Associated Press, high school football for The Washington Post and earthquakes for the Los Angeles Times.” Perhaps you might consider sharing the links to poetry and reporting examples with faculty in your English and journalism departments, just in case they have not yet considered what is available to their students today.
Would you like to interact directly through an API with GPT-3? There are a number of ports online that allow you to engage directly. Some are text-only, such as https://blenderbot.ai/; others enable you to engage with your own voice speaking with an avatar, such as https://iamsophie.io/. There are many more options to be found in an internet search on GPT-3. The sites I used are not necessarily endorsed by me and are not a comprehensive list.
The tools linked above are available to students today. Some will be using them this semester without attribution. How does the advent of such sophisticated research and writing impact what and how we teach in higher education? As access to these services become as common as electronic calculators, do we need to change the learning outcomes and modes of assessments in our classes? Who is leading consideration of the impact of AI on the curriculum at your university?