Recently, OpenAI has unveiled a new artificial intelligence called ChatGPT Edu developed to improve learning and various processes in university settings. Being built upon OpenAI’s recent GPT-4o, it brought important additions for students and teachers enhancing the performance of the model.
What is ChatGPT Edu?
ChatGPT Edu is an academic version of ChatGPT which is designed for use by universities. Its goal is to help education stakeholders utilize AI sustainably by offering accessible strategies for integrating it into everyday processes. So, it was explained that this model is intended to help students, faculty and staff, researchers, and campus administrative services.
The following are the main features of the application ChatGPT Edu:
Advanced AI Capabilities: ChatGPT Edu utilizes the advanced GPT-4o model that performs incredibly well in reading, coding, and math problems. It has enhanced features such as data manipulation, web access, and abstracting of documents.
Customizable GPTs: Universities can develop their own named GPTs in the tutorial model which are specific for the institutions and they can share it in the campus area.
Enhanced Language Support: Edu has embraced over 50 languages, thereby offering enhanced language in terms of quality and speed.
Higher Message Limits: Compared to the freemium model of ChatGPT, users will be able to compose many more messages per given timeframe.
Robust Security and Privacy: The model applies robust security elements, data protection measures, and other management capabilities, like group permission and SSO. However, it is vital to know that information from ChatGPT Edu conversation will not be incorporated into OpenAI model training.
Universities Embracing AI
Many more marquee universities have now incorporated AI models within their universities with positive outcomes. To be specific, Oxford University, Wharton School at the University of Pennsylvania, University of Texas at Austin, Arizona State University, and Columbia University have incorporated the app called ChatGPT Enterprise to improve different aspects of the academic experience.
Real-World Applications
Personalized Teaching: They explain how machine learning makes it possible for teachers to design lessons that can suit the needs of each learner.
Resume Reviews and Grant Writing: Thus, it helps with what to look for in a resume and how to write a grant proposal.
Grading Assistance: With the help of artificial intelligence, educators utilize it to address the issue of grading and assessment, reducing the time spent on it and increasing effectiveness.
Specific examples include:
While teaching at Columbia University, Professor Nabila El-Bassel employs ChatGPT to assess data taken for overdose intervention in a fairly short time.
For example, at Wharton, Professor Ethan Mollick employs ChatGPT to enrich the educational process during the completion of reflective tasks.
At Arizona State University, Assistant Professor Christiene Reves is currently creating a GPT for practice in the German language, which is capable of giving specific feedback and requiring less time from the faculty.
These applications prove how AI can be instrumental in enhancing the experience of both students and faculty with different academic tasks. ChatGPT Edu plans to maintain this great heritage and equip universities with the best tools for implementing AI responsibly and efficiently.
Can Wikimedia Be Trusted in the Age of Misinformation? How AI is Helping During Elections
The Wikimedia Foundation has adopted major measures to counter misinformation during the 2024 election period by relying on AI technological advancements and human efforts. About 60 countries hold regular elections including the European Union, thus nearly half the world’s population has the right to vote. It is in the same vein, millions of people will be out in search of credible information and Wikipedia becomes their best shot at it.
Tools and Strategies
The Wikimedia community has provided its users with several tools they can use to ensure the authenticity of Wikipedia. The following tools comprise bots, extensions, assisted editing programs, and web applications. Some notable tools are:
Bots: Sulipedia bots such as vandalism bots and automated bots that assist in identifying and undoing vandalism including ClueBot NG and ST47ProxyBot.
Extensions and Gadgets: Applications like Twinkle and LiveRC augment the functionality of the editing as well as the reviewing process for volunteers.
Assisted Editing Programs: Tools such as Huggle, which helps in identifying and correcting mistakes in a short amount of time, or AutoWikiBrowser, which is designed to deliver erroneous edits on the A5 widow.
Web Applications: Examples include Checkwiki, a platform that helps volunteers identify and handle fake news, and CopyPatrol, another platform that assists volunteers in the same cause.
The Role of AI
As mentioned by the anti-disinformation strategy lead of the Wikimedia Foundation, Costanza Sciubba Caniglia, AI is meant to be used as a tool and aid for people. AI programs assist volunteers in such activities that require a lot of time and that virtually could be done mechanically, so that the human editors can be involved in more complex cases. For instance, AI algorithms can detect suspicious behavior that may involve vandalism but it is the discretion of another human being to decide if the changes made should remain or be reversed.
Maintaining Neutrality and Accuracy
Wikipedia’s one of the policies that serve as guidelines is the neutrality policy. This means that all the content that is being used has to be accurate and from credible sources and it also means that the information being provided should not be twisted in any given particular way.
Page Protection: This has been done to allow administrators to lock pages for a while so that newcomers or immature users will not be able to tamper with them.
Watchlists: Specifically, watchlists are designed to inform experienced editors about the changes on the pages of interest, which ensures immediate response to the misinformation.
Arbitration Committee: This group makes sure that certain issues do not lead to edit wars as they are responsible for the formulation of rules governing contentious issues. Every action is recorded and documented; any dispute can lead to an appeal.
Also, more than 140 people are active members of WikiProject Elections and Referendums, a task force that aims at increasing the quality and accuracy of provided data in articles about elections and referendums.
Challenges and Future Plans
A pilot study during the 2020 US Presidential elections saw Wikimedia partner with universities to study where misinformation could infiltrate Wikipedia. This work has facilitated the emergence of new artificial intelligence technologies that assist in the identification of unsourced statements and the work of malicious users. Over 56k members contributed to safeguarding around 2k pages related to the elections, signifying the community’s commitment to keeping false info at bay.
Moving ahead, the Wikimedia Foundation is in the process of considering innovative methods of providing information responsibly. This can range from the utilization of AI in generative methods to improving the production and sharing of knowledge.
Thus, together with the help of artificial intelligence and people’s work, Wikimedia strives to present complete and unbiased data to voters, enough to give them all the necessary information.
Can AI Smell? Scientists Are Cracking the Code
Modern features such as voice commands and face recognition are all supported through the recent sound and lighting technologies within smartphones. But when it comes to the olfactory sense, we have not come across something like this – or something like this – not until now. Recently, AI has approached the serious task of endowing computers with the ability to smell, also known as machine olfaction.
The Challenge of Smell
Over a century ago, Alexander Graham Bell actually promoted the development of a science of olfaction similar to the science of audio or vision. While the sense of sight depends only on several cells of the eyes, the sense of smell includes about 400 various types of receptor cells in the nose.
How Machine Olfaction Works
Some of the key components of machine olfaction include the sensors that detect and recognize airborne molecules – which is similar to the way our olfactory receptors do. However, for these sensors to be of any utility, they should be able to know what these molecules smell like to the human nostrils. This is where machine learning comes in enabling the solution to be acquired through computer means.
In this context, the concept of machine learning is quite important.
Machine learning, specifically deep learning has enabled major breakthroughs that have already been seen in regard to technologies like voice assistants and facial recognition. For machine olfaction, technologies assist in relaying details of odor-causing compounds to general terms; sweet or fruity, for example To achieve this, huge data sets are required to train these AI systems.
The Breakthroughs
A similar contest called the DREAM Olfaction Prediction Challenge in 2015 only offered the data to train models to predict smells having molecular structures. The accuracy of the best individual models was published in 2017 using a method known as random forest, which is a collective name for numerous decision trees.
Recent Advances
Thus, after the DREAM challenge was completed, further developments were made in machine olfaction. When COVID was experiencing its peak, many individuals lost their ability to smell which made this sense more recognized. Such initiatives as the Pyrfume Project have made larger datasets accessible to researchers increasing the rate of research.
And by 2019, datasets have expanded vastly and Google Research takes a type of deep learning called graph neural networks to the state-of-art in machine olfaction. Their model established the ‘principal odor map’ where related smells were aggregated despite the fact that the molecules involved are quite diverse.
Promising Applications
The recent breakthroughs in the field of machine olfaction open up a myriad of possibilities as to its use such as designing perfect fragrances, upgrading insect repellents, fashioning new chemical interfaces, diagnosing diseases in their early stage, and enriching virtual reality environments.
Based on the preceding arguments the future of machine olfaction is bright and the future holds a lot of possibilities in store for this form of technology especially in the field of interaction with the world by experiencing it with smell.
