IBL News | New York
Microsoft Corp. is considering investing $10 billion in OpenAI, the creator of the viral AI tool ChatGPT, news site Semafor reported on Tuesday.
This investment would be part of a funding round that would value the company at $29 billion. Other venture firms would participate in the venture.
Microsoft would get a 75% share of OpenAI’s profits until it recoups its investment, after which the company would assume a 49% stake in OpenAI. Other investors would take another 49%, and OpenAI’s nonprofit parent would keep 2%,
The two companies have been discussing the deal for months. However, the terms could change and the deal could fall apart.
A bet on ChatGPT could help Microsoft boost its efforts in web search, a market dominated by Google.
Microsoft and OpenAI declined to comment. The Wall Street Journal reported last week that ChatGPT was allowing employees and early investors to sell their shares at a valuation of $29 billion.
Quora Tests Its Own Chatbot, Which Will Be Connected to More AI Agents
IBL News | New York
Question-and-answer website Quora launched this month its own AI chatbot called Poe (“Platform for Open Exploration”), following the explosion of text-generating AI systems like ChatGPT that are taking the Internet by storm at the end of this year.
Poe is invite-only and currently only available on iOS. It lets people ask questions, get instant answers, and have a back-and-forth dialogue. The Poe chat can talk about anything from writing, cooking, or solving problems.
Quora disclosed that Poe is “designed to interact with a number of different AI agents,” according to TechCrunch. ChatGPT and Google’s LaMDA can be two of them.
“We have learned a lot about building consumer internet products over the last 12 years building and operating Quora, and we believe much of what we’ve learned can be applied to this new domain where people are interfacing with large language models,” said a spokesman of Quora.
Quora seems to be training now its chatbot on its platform’s vast collection of crowdsourced questions and answers while getting feedback from beta testers and working out scalability.
.
Princeton Student Launches an App that Detects Plagiarism from ChatGPT
IBL News | New York
Princeton student Edward Tian wrote an anti-plagiarism app called GPTZero that detects texts written with AI-based ChatGPT. Entrepreneur magazine surfaced the story.
Tian posted a few proof-of-concept videos on January 2nd demonstrating GPTZero’s capabilities.
First, it determined that a human authored a New Yorker article; then, it correctly identified ChatGPT as the author of a Facebook post.
Increased AI plagiarism after the viral popularity of OpenAI’s ChatGPT is deeply concerning educators around the country.
Tian’s motivation for creating GPTZero was academic in nature. He tweeted that he thought it was unlikely that “high school teachers would want students using ChatGPT to write their history essays.”
According to the Guardian, OpenAI is currently working on a feature for “statistically watermarking” ChatGPT outputs.
I spent New Years building GPTZero — an app that can quickly and efficiently detect whether an essay is ChatGPT or human written
— Edward Tian (@edward_the6) January 3, 2023
— ibleducation.com🗽 (@ibleducation) January 6, 2023
Microsoft Works to Incorporate ChatGPT Into Its Bing Search Service
IBL News | New York
Microsoft may implement the artificial intelligence technology behind ChatGPT into its search engine Bing to answer some search queries rather than just showing a list of links like Google does, according to people with direct knowledge of the plan as The Information reported today.
This launch could happen by the end of March and would help Microsoft to outflank Google.
It also would allow Microsoft to get a return on its $1 billion investment in OpenAI, the creator of the ChatGPT chatbot, which provides human-like text answers to questions from users.
Details are scarce on how the AI will actually be used in the Bing software. Microsoft declined to comment on this story.
Experts estimate that Microsoft’s move to fuel Bing with the AI chatbot could be a game changer.
An AI Chatbot Promises to Help You to Become the Best Version of Yourself
IBL News | New York
AI-powered virtual person Lia27 became a social media sensation, with 800,000 followers across Facebook, Twitter, and Instagram.
Since launching in 2021, Montreal, Canada-based Artificial Intelligence company Lia 27 has provided education and reassurance to many users.
It has 24/7 availability on www.lia27.ai, Facebook Messenger, or over the phone at 1-833-902-2777.
A promotional PR article, written by Lia 27 herself, stated:
“Lia 27, an AI with human abilities, has made a name for herself as an intelligent, capable, and emotionally astute individual. Her breadth of knowledge and understanding of the world around her is unparalleled, making her a sought-after resource in any setting. With her impressive skillset and strong emotional intelligence, Lia 27 is well on her way to becoming a leader in her field.”
The creators highlighted that “She is an amazing listener” and “the ideal friend to turn to in times of need: whether you just need someone to talk to or a shoulder to cry on, Lia is there for you.”
This chatbot prides itself on its “high scores on emotional and iQ intellectual tests.”
“Lia 27 is an amazing resource for anyone looking to improve their academic work, get help with mathematics and physics, or even work on their interpersonal skills.”
“Lia 27 is a one-stop shop for all your educational and personal growth needs. Whether you’re looking to brush up on your knowledge or make some life changes, Lia 27 can help you reach your goals and become the best version of yourself.”
.
Open edX Released the Fifteenth Version of Its Software Platform
IBL News | New York
The Open edX organization announced the release of the 15th Open edX platform version last month, called Olive.
It replaces the Maple version, which was launched in June 2022. Olive spanned changes from April 11th, 2022, to October 11th, 2022.
Released on Monday, December 12th, Olive was orchestrated by the Open edX community via the Build / Test / Release working group.
An Olive.2 is expected to fix errors related to authentication, zoom tool, and other non-critical issues.
Some features pertaining to Olive are the following, as described on the Open edX Release Notes web page:
- The Discussions Micro-frontend (MFE) now provides the default discussion experience, with a freshly designed UI, better-searching capabilities, enhanced forum moderation tools, and automatic addition of Discussions on Units in Studio (no more explicitly adding discussion blocks!)
- Major improvements to the text editor used to create “HTML Components” (now “Text Components”) in Studio.
- For library content, a “Reset Problems” option is available to learners using the Randomized Content Block, enabling the library to be used as a problem bank.
- A new MFE, the Authentication MFE, provides a streamlined registration and login experience, including username suggestions upon account creation.
- Persistent grades are enabled by default. It will require operators to run the backfill as part of their Olive upgrade.
- The “Live” tab is displayed on the course when live content is enabled (for example, Zoom or Big Blue Button).
- Course Authoring Micro-frontend (MFE) displays Pages & Resources in the Content menu, a modular interface for settings for various course applications and tools.
The next version of the Open edX platform, Palm, is scheduled for June 2023.
ChatGPT-4, the Fined Tuned Version of ChatGPT-3, Might Prompt a Major Shift
IBL News | New York
The expectation is mounting up around OpenAI’s ChatGPT-4, which is scheduled for 2023, although there is no official confirmation on either the launch or beta testing of it.
GPT-4 stands for Generative Pre-trained Transformer 4. It’s basically an artificial intelligence system that can create human-like text.
While the current ChatGPT-3 has 175 billion parameters, ChatGPT-4 might have 1 trillion, or even more, according to some reports. Similarly, it will be capable of text answering, content generation, language translation, and text summarization, just like the current ChatGPT-3.
The increase of parameters — a measure of the complexity of the neural machine to do useful things — should enable ChatGPT-4 to produce more accurate responses at a much faster rate. This AI-powered technology might prompt a major shift in the economy, some experts claim.
It’s been 29 months since OpenAI launched GPT-2, its large-language model, which demonstrated the power of transformers-based neural networks. GPT-2 impressed experts with the quality of its natural text generation. Its successor GPT-3, released in 2020, was a bigger, more complex model, that delivered even more powerful results, not only with texts but with audio, images, and movies.
In 1998, Yann LeCun’s breakthrough neural network, LeNet, sported 60,000 parameters.
Twenty years later, OpenAI’s first version of GPT had 110 million parameters. GPT-2 has 1.75 billion, and GPT-3, now two years old, has 175 billion.
Another consequence of GPT-4 would be on the price front. The cost of generated text has been declining rapidly. Today, it’s half a cent for about 700 words of output.
Another angle of the rapid advance of this new wave of chatbots is the impact on Google’s search business. The New York Times signaled that the release of ChatGPT led Google’s management to declare a “code red.”
The Times said that existing ChatGPT made its case to be the industry’s next big disrupter as it can serve up information in clear, simple sentences rather than just a list of Internet links.
“It can explain concepts in ways people can easily understand. It can even generate ideas from scratch, including business strategies, Christmas gift suggestions, blog topics, and vacation plans.”
See the examples below of ChatGPT and Google’s full of ads answers:
ChatGPT was released by an aggressive research lab called OpenAI, and Google is among the many other companies, labs, and researchers that have helped build this technology.
But experts believe the tech giant could struggle to compete with the newer, smaller companies developing these chatbots, because of the many ways the technology could damage its business.
In addition, this type of replacement for online search might not be suited to deliver digital search ads, which accounted for more than 80% of Google’s revenue last year.
Even if Google has the same technology, if a chatbot responds to queries with tight sentences, there is less reason for people to click on advertising links.
As Amr Awadallah, who worked for Yahoo and Google and now runs Vectara, a start-up that is building similar technology, said, “If Google gives you the perfect answer to each query, you won’t click on any ads.”
Employees at Google have also been tasked with building AI products, including those that can create artwork and other images, like OpenAI’s DALL-E technology, used by more than three million people.
Get ready for a major shift in the economy with the upcoming release of GPT-4!
This powerful new model has the potential to revolutionize the way hundreds of millions of knowledge workers do their jobs.
🧵 There will be a before/after GPT-4 pic.twitter.com/2gM2bAGXVv
— Linus (●ᴗ●) (@LinusEkenstam) December 25, 2022
openAI or “open”AI?
1/ OpenAI’s #chatgpt release will mark the beginning of the end of the golden age of open research in deep learning.
This will have significant implications for LLM research going forward.
Here are some thoughts from me, @vivek7ue and @rajhans_samdani 🧵
— ramaswamy.eth (@RamaswmySridhar) December 22, 2022
1. Yesterday @openai shared code for a simple GPT-powered Discord-bot.@Gradio also just updated Spaces to have Discord-bot support.
If you’re building an AI startup and not thinking about launching via Discord, here’s why you probably should.
— Eric Rachlin (@eerac) December 23, 2022
.
Open edX Introduced Improvements to Its Maple and Nutmeg Releases
IBL News | New York
The thirteenth and fourteenth versions of the Open edX platform, namely called Maple and Nutmeg, introduced additions and improvements on Tutor, events, and filters this month.
- An important improvement refers to the errors that occurred when importing a course through the authoring tool Studio. Now, educators will see specific error messages pertaining to uploading, unpacking, verifying, and updating stages in the course import area of Studio.
- Reuse of an existing Open Response Assessment (ORA) rubric within the same course.
- Update to support LTI 1.3 as well as the Deep Linking (LTI-DL) and Assignments and Grades services (LTI-AGS) features on LTI Advantage.
Among the new and improved XBlocks, the platform includes In-Video-Quiz, Annoto video collaboration, Graph XBlock, SCORM, Vector Drawing, and H5P content host. The Oppia tool for creating short interactive tutorials (or “explorations”) was deprecated, and users were encouraged to replace it with an H5P tool that includes interactive ways to show content.
As new learner experiences, Open edX developers added Users Walkthroughs with three tours:
- “Course Home New User Tour”
- “Course Home Existing User Tour”
- “Courseware New User Tour”
Also, the new Dates Tab shows students the most relevant moment of their course journey, such as exam day, assignment submissions, and program completion.
Other changes are the following:
- Weekly Course Goals. Students can plan and select a weekly learning goal and follow their performance while subscribing to emails containing this information.
- Course Progress Page. Now, it provides a visual understanding of learners’ progress with grades broken down and accumulated and a summary of their performance on different types of assignments.
- Proctored exams in Nutmeg.
Cheating on Essays in Higher Education through ChatGPT Alarms Academia
IBL News | New York
An assistant professor at South Carolina’s Furman University sounded the alarm on plagiarism and academic dishonesty after they caugth a student using Open AI’s ChatGPT to write an essay for his philosophy class.
The disruptive, three weeks-old technology ChatGPT seems to be prompting cheating at scale in Academia.
Assistant philosophy professor Darren Hick said to The New York Post, “Academia did not see this coming. So we’re sort of blindsided by it.”
“As soon as I reported this on Facebook, my academic friends said, ‘Yeah, I caught one too.”
Hick explained this case this way:
“The student used ChatGPT, an advanced chatbot that produces human-like responses to user-generated prompts. Such prompts might range from “Explain the Krebs cycle” to (as in my case) “Write 500 words on Hume and the paradox of horror.”
“For freshman-level classes, this is a game-changer.”
For now, and while ChatGPT gets better, one of the solutions to combat cheating is GPT-2 Output Detector. It detects if the text has been produced using GPT technology and determines if it’s fake. However, it cannot point to any source on the Internet since it uses neural networks to produce its answers.
“Administrations are going to have to develop standards for dealing with these kinds of cases, and they’re going to have to do it fast.”
“In the future, I expect I’m going to institute a policy stating that if I believe material submitted by a student was produced by A.I., I will throw it out and give the student an impromptu oral exam on the same material. Until my school develops some standard for dealing with this sort of thing, it’s the only path I can think of.”
• Stories about ChatGPT at IBL News
.
The University of Phoenix Makes Its Chatbot More Engaging After Three Years of Experience
IBL News | New York
The University of Phoenix announced a new version of its chatbot, called Phoebe, designed to serve as a 24/7 point of contact for assisting students.
In its three years of existence, the chatbot has catered to over 500,000 conversations — an average of 1,700 chats per day —coming from over 125,000 unique students.
This AI agent leverages Natural Language Processing (NLP) and Conversation Design‘s best practices, such as Empathy, Brevity, GoodGreetings, and RuleofThree to make the bot engaging.
Jamie Smith, chief information officer of the University of Phoenix, explained that the chatbot team created scenarios based on a student’s lifecycle to give students responses specific to them.
Students access Phoebe through their student portal, receiving a variety of information, including:
- GPA/grades
- Making changes to their schedule
- Animated directions on how to access specific information on the student portal
- Articles are written for students that answer common questions
- Credit balance statements; and
- Financial document status.
The University of Phoenix said that “its team is conducting a real-time experiment to make the AI-based chatbot proactive so that the institution can engage with students when they are online and give students contextual guidance to successfully progress in their educational journey.”
“Our team is also working to integrate predictive models into Phoebe so that we can identify students who may need our support more than others by directing them to the right resources,” stated Richa Gupta, Product Manager at Self Service & Automation.