Generative AI and Language Processing🤖✍️[HTC #52]
ChatGPT is a real conversation starter (ChatGPT came up with this)
Welcome to the 52nd edition of Hold the Code! In this edition, each article will cover the growing industry of generative AI and its implications on our everyday lives. Make sure to share HTC with a friend and as always, Happy Reading!
The Rise of Generative AI
Image Source: Search Engine Journal
Generative artificial intelligence – AI that can take prompts and output text, images, or sounds – has taken Silicon Valley investors by storm. But how much of this excitement is grounded in reality?
The most popular development in the field came just at the end of last year, when artificial intelligence lab OpenAI released ChatGPT, the most powerful publicly available chatbot to date. The tool quickly became a sensation, with millions of people using it for tasks ranging from answering questions to writing fully-fledged essays. This new technology – the product of over a decade of research – could potentially revolutionize the way we interact with everything from image editors to digital assistants to search engines.
A new wave of investments
The excitement surrounding generative AI has spurred major investments in the area. OpenAI is currently in talks to complete a deal that would value it at around $29 billion, over double its valuation in 2021. It’s estimated that generative AI companies received over $1.37 billion from investors in 2022 across 78 deals – almost as much as the investments from the previous five years combined.
“It’s the new ‘mobile’ kind of paradigm shift that we’ve been all waiting for,” said Niko Bonatsos, an investor at venture firm General Catalyst. “Maybe bigger, too.”
Hype or reality?
However, some worry that the buzz around generative AI has gotten ahead of reality.
Michael Dempsey, investor at venture firm Compound, notes that the surge of investor optimism may be influenced partly by the downturn that the tech industry has faced over the past year, which included a crypto crash, poor-performing stocks, and mass layoffs.
The technology has also raised ethical questions about how copyright laws factor into generative AI. For instance, companies still do not have a clear answer of whether they should need permission to use the data that trains their algorithms.
Despite these concerns, interest in generative AI remains on an upward trend and doesn’t show signs of slowing anytime soon.
“For four and a half years, people thought we were nuts,” said Jordan Jacobs, a partner at Radical Ventures, a firm created five years ago specifically to invest in AI. “Now, for the past six months, they’ve thought we were geniuses.”
Generative AI in Education
Image Source: Dalle/The74
ChatGPT has become an increasing cause of worry among educators and school instructors. Due to concerns about plagiarism, as well as worry about response inaccuracy, school districts across the United States have begun taking action to try and prevent student usage of the AI-powered tool.
How does ChatGPT work?
Before addressing the potential pedagogical implications of this technology, it is important to review how ChatGPT generates language in a conversational style. The chatbot is built atop OpenAI’s Generative Pre-trained Transformer 3 (or, “GPT-3” for short), which is 175 billion parameters in size and trained on over 10 billion words. OpenAI used supervised and reinforcement learning in order to “fine-tune” the chatbot, which included the usage of humans to improve model performance. By ranking the different responses ChatGPT gave to commands, developers employed a type of Proximal Policy Optimization to reward the bot for accurate, convincing answers.
How is it affecting students?
Though intended for use as a conversational chatbot, ChatGPT quickly gained fame, (and notoriety), for its ability to produce seemingly genuine responses to student essay prompts. Students on social media quickly spread word about the methods for which ChatGPT technology could be employed, including essay responses, discussion posts, and automated feedback on written responses.
Evidently, concerns have been raised about the ethics of using ChatGPT to generate assignment responses. Although “detection bots” for ChatGPT are being developed, it is unclear if these tools can be effectively used to detect AI language model usage, or if ChatGPT can simply be further refined to avoid detection. In addition, teachers have begun debating whether students should be allowed to use ChatGPT in their writing.
How are schools responding?
In New York City, agency officials confirmed that students and teachers would no longer be able to access ChatGPT technology. Citing “negative impacts” on learning and potential concerns regarding the “safety and accuracy of content,” the New York City school network blocked access to the website.
“...while the tool may be able to provide quick and easy answers to questions, it does not build critical-thinking and problem-solving skills, which are essential for academic and lifelong success.” said Jenna Lyle, an Education Department spokesperson.
Irrespective of the school network ban, students can still circumvent the block by using a VPN or accessing the service from a personal device at home. Regardless of whether educators choose to stand by their ban on ChatGPT, or if AI tools slowly become integrated into the educational curriculum, the ChatGPT represents a fundamental shift in how educators will grapple with technology moving forward.
Copyright and Generative AI
Image Source: Nature
Can you tell what articles are written by a person? Have you ever considered that humans may not have written them?
Now, computers play an active role in every aspect of writing, for instance, spell check and suggestive editing. As a result, the line between inspiration and plagiarism is thinned further. Bruce Weinstein, an ethicist writer for Forbes magazine, expresses a few of his insights and experiences with AI and writing, which are summarized below.
The prompt
Weinstein gave ChatGPT the following prompt, while also answering it himself:
In 200 words or less, what are three things that get in the way of doing the right thing?
ChatGPT’s response was shockingly accurate, though it lacked the anecdotal nature of Weinstein’s answer. ChatGPT’s gave this as the full response:
The problem
For language processing algorithms like ChatGPT, much of the source material that is used to create their prompts is “inspired” from a large corpus of writing, making plagiarism a key issue. Additionally, some writers could incorporate responses from a ChatGPT reply and not cite.
“Artificial intelligence can also be used for ethically troubling purposes, such as writing sections of articles or entire books for us, which we then claim as our own.”, quoted Weinstein.
As ChatGPT and other Generative AI become more advanced and involved in our daily lives, one question remains: How can we ethically use Generative AI?
Have any thoughts or questions about generative AI? We’d love to hear them! Please send anything you’d like to share through our Google Forms here.
Love HTC?❤️
Follow RAISO (our parent org) on social media for more updates, discussions, and events!
Instagram: @Raisogram
Twitter: @raisotweets
RAISO Website:
https://www.raiso.org