• Work With AI
  • Posts
  • OpenAI VS. Plagiarism: How Will The Text Classifier Fair?

OpenAI VS. Plagiarism: How Will The Text Classifier Fair?

02-01-2023

⚡ Today’s Highlights

  • 📰 News: OpenAI is trying to catch AI plagiarism and the DoD is trying to prevent “The Terminator” from becoming reality

  • 🦾 Tools: News aggregation and better data processing

  • 🧠 Resources: OpenAI shares a brief overview of ChatGPT for educators

  • 📅 Events: The World AI Cannes Festival, RE•WORK AI Summit West, Data Science Salon Austin

📰 Today's Top Stories

(6 min read) (Source: OpenAI)

TLDR: OpenAI has developed an AI Text Classifier that can (attempt to) determine if a text was written by a human or an AI. However, it's not perfect and has limitations, such as being unreliable on short texts, misidentifying human-written text, struggling with non-English languages, and differentiating between facts written by humans versus AI.

The tool can be used to detect plagiarism, AI-powered disinformation, and other misuses. But, beware, the tool can be easily evaded by editing the AI-written text. Schools are grappling with the impact of ChatGPT, with some banning its use, while others are embracing it as a tool to enhance critical thinking skills for both students and teachers. Check out the announcement and discussion on Twitter below.

The Big Picture: OpenAI is leading the charge against generative AI being used for nefarious purposes. CEO Sam Altman stated previously that, at the moment, it is nearly impossible to detect AI plagiarism. Many different companies are attempting to create tech that can detect AI plagiarism such as those experimenting with the use of a “watermark” attached to AI-generated content. This detector presents an excellent start in the battle against those pesky plagiarists.

(4 min read) (Source: Popular Mechanics)

TLDR: The Department of Defense has updated its guidelines for using autonomous and AI-led weapons (DoD Directive 3000.09), ensuring that human judgment is still required in their deployment. The UN has also been discussing the topic of autonomous weapons and the level of human control needed since 2018.

The DoD says the use of AI in weapons is still acceptable as long as it aligns with their AI Ethical Principles and Responsible AI Strategy. The goal is to minimize the risk of unintended engagements and ensure compliance with the law of war, treaties, weapon system safety rules, and rules of engagement. The future of AI in weapons and its impact on geopolitics is an ongoing discussion that will shape the next decade.

The Big Picture: If you’re reading this right now, you’ve probably seen “The Terminator” and know all about Skynet, the world-ending AI that started it all. In our world, we may still have some time before chiselled time travelling robots that look like Mr. Olympia become the weapon of choice. However, the recent developments in AI will likely mean that military applications for AI are just around the corner. It is critical that legislation and safeguards for this tech are implemented worldwide in consideration of growing political conflicts around the world.

(10 min read) (Source: The Atlantic)

TLDR: ChatGPT may sound intelligent, but don't be fooled! A recent study by cognitive scientists and linguists suggests that language is just a messenger, not the medium of thought. The human brain has a network of neurons that operates independently of language, responsible for activities such as math, music, and coding.

Patients with aphasia, a loss of language ability, can still perform these non-linguistic tasks, supporting the idea that language is separate from thought. AI language models like ChatGPT may be good at stringing words together but struggle with understanding the human nuances of context, social cues, tone, and more. The chatbots are still working on it...for now.

The Big Picture: Chatbots are getting very good at doing technical tasks like coding and writing essays, but they aren’t quite able to convey some of the intricacies of human communication. This provides some hope for human creativity. They’ve attracted substantial controversy in the art world but it seems that they might not be able to create a masterpiece born from the complexity of human experience.

The stories, songs, artworks, movies, etc. that people consider to be once-in-a-lifetime creations may retain their integrity because of the people that created them. For now, AI is just tainting the creative process for some folks.

(7 min read) (Source: Forbes)

TLDR: Fintech is revolutionizing the financial world with AI tools like RPA (robotic process automation), ML, and AI. These 'augmented intelligence' systems are enhancing human decision-making and providing a more complete understanding of customer behaviour while removing bias.

From application to customer support, these tools help tailor products for subprime borrowers, reducing the risk of missed payments and overdrafts by anticipating customer needs and proactively providing support, lowering the barrier to entry and success. The goal of AI in lending is to support fairer outcomes and improve financial well-being for customers.

The Big Picture: Hopefully the introduction of AI in finance will help reduce some of the biases that have hindered people from lower-income backgrounds from being successful financially (though there are still some kinks to work out in the biases that are present). It should lend support and make navigating the financial world somewhat easier for regular folks who’d like to take better care of their finances and explore opportunities for growth.

🦾 Trending Tools

  • The co-founders of Instagram just announced Artifact, a news aggregation app that uses an algorithm similar to TikTok, Google Reader, or Twitter to provide a curated news feed. The app is currently in private beta.

  • Glass AI is an experimental new feature of the Glass Health platform that can generate a diagnosis or treatment plan based on a one-line description of a health-related problem. It is NOT meant to be used by the general public but is being shared here as it will likely be used to reduce friction in underfunded healthcare systems.

  • The Rose Web Platform is a data processing and visualization tool. It was designed by finance professionals from some of the world’s biggest hedge funds. It is being presented as a cross between ChatGPT, Excel, and Tableau. Check out the Twitter announcement here.

  • Intercomm, the customer-service business platform is introducing 3 new features based on OpenAI’s ChatGPT to support its B2B toolkit and make the lives of businesses and customers a bit less painful. Similar to ChatGPT, these features include helping customer service reps respond to customers, summarizing interactions, and generating articles for the purposes of documentation.

🌎 Popular Content

2. He’s not the greatest guy but these biases are getting kind of crazy… (Twitter)

3. Hold on a second, what exactly do you mean by that????? (Twitter)

4. Check out this growing list from OthersideAI of the “definitive AI market map”. (Twitter)

5. Is ChatGPT coming for software engineers' jobs? Check out the discussion on Twitter here. (Twitter)

👀 More Reading

🧠 Resources

  • OpenAI shared a brief overview for educators seeking to learn more about the capabilities, limitations, and considerations for using ChatGPT for teaching and learning.

  • Find the videos from DeepLearning AI’s Coursera programs on machine learning as well as recorded events on their YouTube channel. By providing a platform for education and fostering a tight-knit community,DeepLearning AI has become the pathway for anyone looking to build a career in AI.

 📅 Upcoming Events

  1. The World AI Cannes Festival (February 9-11, 2023. Cannes, France + Virtual), where decision-makers and AI innovators meet, where the most promising innovations and technologies get into the spotlight, where those who are currently building the world’s most game-changing AI strategies and use-cases will be on stages.

  2. RE•WORK AI Summit West, Deep Learning Summit (February 15-16, 2023. San Francisco, CA), a chance to hear the latest technology advancements, practical examples of how to apply AI to solve challenges in cross-industry settings, business and society, and delve deeper into the work of leading AI experts in a series of presentations, panel discussions, interviews and fireside chats.

  3. Data Science Salon Austin (February 21-22, 2023. Austin, TX + Virtual) is a two-day 500-person conference focused on AI and machine learning applications in the enterprise. The intimate event curates data science sessions to bring industry leaders and specialists face-to-face to educate each other on innovative new solutions in artificial intelligence, machine learning, predictive analytics and acceptance around best practices.

 💼 Jobs

Powered by AI. Curated and edited by Humans.

How'd you like today's email?

Login or Subscribe to participate in polls.

We'd love to hear from you! Leave us comments or feedback by replying to this email!