ChatGPT from teacher's perspective
Vedran Miletić, Faculty of Informatics and Digital Technologies
Build Your Career Garden – Learning Teaching and Training event, Faculty of Civil Engineering, University of Rijeka, 30th of March 2023
Faculty of Informatics and Digital Technologies
- founded in 2008. as Department of Informatics at University of Rijeka
- became Faculty of Informatics and Digital Technologies in 2022.
- several laboratories and research groups; scientific interests:
- computer vision, pattern recognition
- natural language processing, machine translation
- e-learning, digital transformation
- parallel applications on supercomputers
Group for Applications and Services on Exascale Research Infrastructure (GASERI)
- focus: optimization of computational biochemistry applications for running on modern exascale supercomputers
- vision: high performance algorithms readily available for academic and industrial use in state-of-the-art open-source software
Summary of news regarding ChatGPT over last 4 months
- ChatGPT is not an evolutionary technology development like mobile gaming or expansion in what's possible using web standards
- it is a revolutionary technology development, similar in significance to:
- the Internet
- Google Search
- no more website categories like Yahoo! Directory required
- social networks
- iPhone (and smartphones in general)
- ChatGPT changes how information is retrieved by the user, challenges web search far more than social networks ever did
- started the era of mainstream everyday usage of AI tools
Gradual or instant change?
similar example from the past: desktop to web transition
- 1997. (the pre-Google era) Microsoft Windows is the de facto standard, desktop applications dominate software for PCs
2007. (the Google era) Paul Graham says Microsoft is Dead:
Microsoft's biggest weakness is that they still don't realize how much they suck. They still think they can write software in house. Maybe they can, by the standards of the desktop world. But that world ended a few years ago.
it took roughtly a decade; far from an instant change
similarly, ChatGPT has motivated the gradual introduction of AI-powered technologies into existing tools
Is OpenAI the new Google?
OpenAI was founded in 2015 by Sam Altman and Elon Musk; charter:
OpenAI’s mission is to ensure that artificial general intelligence (AGI)—by which we mean highly autonomous systems that outperform humans at most economically valuable work—benefits all of humanity. We will attempt to directly build safe and beneficial AGI, but will also consider our mission fulfilled if our work aids others to achieve this outcome.
it is known for: Gym (Retro), RoboSumo, Debate Game, Dactyl, DALL-E
What really is ChatGPT, aside from the hype?
- released in November 2022 by OpenAI
- chat based on a large language model; not the first
- uses Reinforcement Learning from Human Feedback
- significant part of model training is performed manually by people
- growth exceeded all expectations
- top 100 prompts
Official website can say
ChatGPT is at capacity right now; static examples:
- List of ChatGPT Examples
- 41 Amazing ChatGPT Demos and Examples that will Blow Your Mind
- 12 Cool Things You Can Do with ChatGPT
- Extra: ChatGPT Examples: 25 Interesting Questions Answered By ChatGPT
The AI wave (1/2)
- Microsoft's new Bing
- 100 million daily active users
- Google's Bard lags behind
- smaller alternative: You.com
- Craiyon/DALL-E mini (similar to OpenAI's DALL-E 2)
- Playform offers a different style
- Bing also has it
- NVIDIA Canvas is similar, but as a desktop application
The AI wave (2/2)
- Notion AI, several features
- similar to Office 365 Copilot, Generative AI in Google Workspace (Docs/GMail)
- Whisper speech-to-text
- specialized tools like BioNeMo, enhanced photo/audio/video editing, prompt-guided 3D modelling etc.
Impact on teaching
- initial resistance, as usual with new technologies
- ChatGPT can grade student solutions to exercies (given the criteria), asses student essays, guide students in improving their writing etc.
- its style of writing is presently recognizable (and therefore hard to hide if the student is lazy), it might not be that way in the future
- Croatia-specific: Croatian language model is not as good as English
Impact on research: Nature
Tools such as ChatGPT threaten transparent science; here are our ground rules for their use:
First, no LLM tool will be accepted as a credited author on a research paper. That is because any attribution of authorship carries with it accountability for the work, and AI tools cannot take such responsibility.
Second, researchers using LLM tools should document this use in the methods or acknowledgements sections. If a paper does not include these sections, the introduction or another appropriate section can be used to document the use of the LLM.
Impact on research: Science
Science Journals: Editorial Policies:
(...) artificial intelligence tools cannot be authors. (...)
Artificial intelligence (AI) policy: Text generated from AI, machine learning, or similar algorithmic tools cannot be used in papers published in Science journals, nor can the accompanying figures, images, or graphics be the products of such tools, without explicit permission from the editors. In addition, an AI program cannot be an author of a Science journal paper. A violation of this policy constitutes scientific misconduct.
- GPT-3 -> GPT-4 transition
- already a part of a released product called ChatGPT+
- used by Bing
- non-English language models likely to improve soon
- within months, maybe a year or two
- remember how well Google Translate performed 10 years ago and compare that to how it performs now
- make sure to visit chat.openai.com and try it
Author: Vedran Miletić