Ryan and Ben are talking about new updates to Google workspace, a wearable AI tech from Humane and false accusations about using AI to write papers. Stay with us as we catch you up on AI news in 10 minutes!
Links:
Google: https://labs.withgoogle.com/
Falsely accused of using ChatGPT: https://goldpenguin.org/blog/falsely-accused-of-using-chatgpt/
S1 Ep55 Google is King for a day
Intro
Hi everybody, this is Ryan and Ben and today we are going to be talking about new updates to Google workspace, a wearable AI tech from Humane and false accusations about using AI to write papers. Stay with us as we catch you up on AI news in 10 minutes!
News
We have decided to cut back to one episode per week coming out every Thursday. We are doing this because as the AI news slows down we still want to fill every one of our episodes with high quality content. So we are choosing quality over quantity. If news starts picking up again we will go back to two episodes a week if needed.
The ChatGPT subreddit has 2,276,000 members
Google's Deepmind CEO says its next algorithm will eclipse ChatGPT (Ill believe it when i see it)
Midjounrey 5.2 has dropped
Toyota Research Institute Unveils New Generative AI Technique for Vehicle Design
Hive Blockchain has moved its massive GPU resources (around 38,000 graphics cards) from mining Ethereum to create and run AI workloads for customers, aiming to create a significant new revenue stream by shifting to the latest trend in the computing industry.
Topic
Google Workspce
Google Workspace now has a waitlist, and we just gained access. It's pretty simple to use, and let me tell you, I'm very interested in your opinion too. Basically, it's like having Bard inside your documents. You can create your own phrases or highlight a paragraph and choose from a set of options: Formalize, Shorten, Elaborate, Rephrase, and even a custom option, similar to Bard or ChatGPT. I tested it out, and so far, I've been pretty impressed. It feels better than Bard, although I'm sure it's somehow linked to it. Overall, this is how Google gains an advantage over ChatGPT in everyday life because you don't have to open a new window. You can do it right there in Docs without clicking out of it. Currently, I only have access to Docs, not Sheets. It seems that you'll be auto-accepted into Docs, but not necessarily into Sheets. In my opinion, this is a significant blow to Microsoft. They showcased their copilot for Word and Excel, and we got excited about the integration into our everyday lives. Then Google comes along and says, "Here, check out this access." I believe it was a missed opportunity for Microsoft, and I think Google attacked hard and came out on top with this. Especially because it's super easy to use, not clunky at all. Again, just my opinion.
Humane AI
Humane, the company founded by ex-Apple employees, known for its ambitious vision of an AI-driven future beyond smartphones, made an exciting announcement recently. Their inaugural device, the Humane Ai Pin, is set to launch later this year, powered by an advanced Snapdragon platform in collaboration with Qualcomm.
This Ai Pin is a wearable device that sits in the front chest pocket of a jacket or a shirt. It has a camera, a projector, microphone and speaker. It functions similarly to an iphone except it does not have a screen. It exclusively uses voice commands to receive tasks. The last glimpse we had of this gadget was during the TED conference in April. Co-founder Imran Chaudhri took the stage and showcased a device demonstrating its capabilities as a voice assistant, making phone calls, providing automated daily summaries, capturing images to retrieve nutritional information, and even projecting a miniature green screen into his hand. The demo video had some flaws and really made me think some of the responses were just canned responses. For instance, he said a phrase and then the Pin repeated that phrase in his voice in French…. The catch is that he never asked Pin to translate the phrase into French. Overall the Pin appears to be more of a gimmick AI tech than something that is actually going to creep into our daily lives.
False accusations
The use of AI detection tools in academic institutions has come under scrutiny due to cases like that of Louise Stivers and William Quarterman. Louise, a political science student at the University of California Davis, was falsely accused of plagiarism by the Turnitin tool. The accusation not only caused immense stress but also had a detrimental impact on her academic performance and mental well-being. In her last semester, Louise had to navigate the challenging task of defending herself while juggling her studies and law school applications. Similarly, William Quarterman faced a false accusation of plagiarism based on the analysis of the AI-detection tool GPTZero, which led to him failing the assignment. The initial accusations placed both students in a position of uncertainty and required them to fight for their innocence.
What makes these cases particularly troubling is that the tools meant to uphold academic integrity ended up causing distress and distractions for innocent students. Turnitin, which was still in beta testing during Louise's incident, claimed a 98% accuracy rate but acknowledged the presence of false positives. Even OpenAI's ChatGPT, acknowledged by its creators to be unreliable in distinguishing between human-written and AI-generated content, raises concerns about the accuracy of AI detection tools. Another alarming incident occurred at Texas A&M University, where Dr. Jared Mumm, an instructor, allegedly misused AI detection and wrongly informed numerous students that they would receive zeros on their assignments, assuming they were written by an AI chatbot. These incidents highlight the gaps and limitations in the deployment and utilization of AI tools in academic settings, raising questions about the undue reliance on imperfect technology and the lack of critical human judgment.
The experiences of Louise Stivers and William Quarterman serve as powerful examples of the potential harm caused by false accusations resulting from AI detection tools. They underscore the urgent need for academic institutions to reevaluate their approaches, strike a balance between technology and human discernment, and ensure fairness and accuracy when dealing with academic integrity cases.
And to top this out, I tested GPTzero and turnitin and have seen similar situations where it was a false positive and then if an AI wrote it it sometimes would say it was human written. I hope professors aren't completely relying on these systems because their 98% correct statement I think is more of a sales point than anything. Also I did testing on 4 different ones and sometimes you'd get 2 saying it was AI and the other 2 saying it was human. Stay safe out there y'all
Closing
Thank you for giving us a listen. Rate and review and subscribe to this podcast so you won't miss our 2 new episodes every week on Monday and Thursday. Give us a follow on LinkedIn and Twitter @ChatGPTReport and check out our website at thechatgptreport.com where you can find links as well as transcripts to each show. Please send us an email at thechatgptreport@gmail.com. We hope you have a great rest of your day.
