This piece was first published on The India Cable – a premium newsletter from The Wire & Galileo Ideas – and has been republished here. To subscribe to The India Cable, click here.
OpenAI is about to release an API (application programming interface) to ChatGPT3, the chatbot which has been getting almost as much public interest as Putin memes since it started batting the breeze at the end of November, and replying to people’s queries with sophisticated essays and artwork. Yesterday, it passed US medical certification tests and Wharton’s MBA test. On the other hand, ChatGPT3 lies, too. It has learned from internet resources and sometimes knows no better than the alumni of WhatsApp University.
ChatGPT3 has become a media sensation and shaken up Silicon Valley. It’s even shaken and stirred Google, which was an early investor in AI, because a conversational AI threatens its search dominance. The current pink slip bloodbath at the search leader has been passed off as rightsizing after a hiring spree during the pandemic, but it accompanied a clear pivot to AI.
But the impending ChatGPT3 API release is not attracting much attention outside the technology press. The lack of interest is puzzling, because the API will, quite literally, open the doors of perception for everyone. An API is like a coupling which apps use to plug into an internet resource. OpenAI will monetise it, which means that developers who pay for it can plug all the logical and creative power of ChatGPT3 into their existing applications, whether it’s Microsoft Word or a video editor. In fact, Microsoft has already contracted access for kits for enterprise customers, who can use AI functions for their business processes.
What does this mean for regular working stiffs? It’s going to slash back office functions in areas ranging from consultancy to programming ― which are important revenue streams for Indian companies. Both involve the fairly mechanical production of documents (computer programs start life as documents written in a text editor, before they are compiled in a form that machines can read). Business contracts, project proposals and wills adapt standardised formats to particular cases. A phone application and dynamic websites like Facebook or Google are also written up in a fairly predictable manner in languages which generally include PHP, Java, C++ and Python. Composing both documents and programs/applications involves grunt labour, which can be done in seconds by a machine like ChatGPT3.
Also Read: The Real Threat From AI Isn’t Superintelligence. It’s Gullibility.
Within a month of the public release of the chat interface, managers who had jacked in were declaring that it had changed their perception of work, and that there would be staffing changes ― a polite way of saying that some human roles would die. One worker revealed that the company which contracted him to compose documents had limited his role. Now, ChatGPT3 would write the docs and he would only be paid to edit out telltale signs of AI-generated text.
But perhaps the most interesting effects will be seen not in the workplace, but in education. Schoolteachers fear the worst, as they fail to distinguish between essays written by ChatGPT3 and human students. Fifty years ago, calculators were banned in high school exams, because that was perceived to be cheating. But slide rules were okay ― that was about logarithms and antilogs, which assured teachers that students had worked their way up at least to Napier.
Now, a broad-spectrum calculator on steroids is within the reach of school students. Educationists may go with the flow, and use AI as an opportunity to rethink outdated teaching methods. For instance, is spelling really important when historically, in many languages, it was not standardised? Are writing skills in a foreign language important (English is, to most of us)? Won’t a machine translation of our thoughts serve the purpose? Is that cheating with a calculator that can write in english, or is it like using a slide rule?
For years, it was anticipated that public access AI would trigger ethical questions. Now that ChatGPT3 has jumped the gun, we need answers quickly. And the answers for countries like India, where inequality is embedded in language, learning and culture, will be very special.
This article was first published on The Wire.