Reality of ChatGPT - Why is Italy banning ChatGPT? | Open AI
233Please find the link to the Management Consulting Program that I personally lead at Blue Chapter! You receive personal feedback every week & work on an international live strategy project. This is not a mock project & our students proudly speak about this live experience in their interviews:)
https://bluechapter.com/consulting-experience-program-gen/
If you are applying for fall 2023/2024 MBA programs, check out the global MBA fellowship program:
https://bluechapter.com/global-mba-fellow/
****
Follow me here
LinkediIn - https://www.linkedin.com/in/pavan-sathiraju/
Instagram - https://www.instagram.com/pavan.sathiraju
****
Will ChatGPT get banned? ChatGPT banned in Italy news is making headlines for some time now. It is expected that Germany too might ban ChatGPT on similar grounds of privacy violation. What is ChatGPT? Why is ChatGPT banned in Italy? Is Germany going to ban ChatGPT too? In this video, I’ve covered what ChatGPT is and what technology it is based upon. After that I talked about ChatGPT issues such as ChatGPT privacy concerns, ChatGPT Data Privacy issues, ChatGPT ethical issues, is ChatGPT biased, and lastly ChatGPT legal issues.
This video will help you understand what ChatGPT is based upon and how AI can be biased. ChatGPT is based on the Transformers model which was developed in 2017. The earlier developed AI models such as sequential neural networks considered the last typed word to predict the upcoming word. This generative artificial intelligence model was based on sequential information processing. In this, the last word was given the most importance. So, if you were to read a whole paragraph then the first sentence in the paragraph would soon lose its relevance. Besides, the first word in a line would have the least importance.
Because of this, the sequential information processing-based AI models couldn’t write a good book summary. Transformers are the AI model that came into existence to tackle this problem. It considered the whole sentence as an input rather than just the last word. In this way, the transformers model used the whole sentence to predict the set of next words. The transformers model didn’t give weight to specific words in the sentence. That’s why ChatGPT can synthesize articles and summarize books. ChatGPT took information from all over the web. These data points are taken from all over the web to predict what would be the next set of words.
But this required terabytes of data processing. The answers were still not in the conversation style. To achieve this, actual humans took the human and ChatGPT role and developed tons of conversations that would inform the AI model on how to converse like a human being. In this way, the model was able to imitate this model clearly. Now after taking all the information from the web and adding the conversational style, the ChatGPT team added the boundary conditions. Such as if someone asks ChatGPT for ways to commit suicide or rob a bank then ChatGPT wouldn’t answer.
After this, we get to the issues with ChatGPT. The first such issue is ChatGPT being biased. Many societies around the world don’t even have access to the internet which means their inputs are not considered by ChatGPT. Besides, ChatGPT and most of the AI models are developed by white males. A Google Researcher who brought up this problem was fired. This is a problem because the marginalized groups which are not developing AI models are at a disadvantage in this way.
The second issue is that ChatGPT does not have current data which means anything after 2021. Training the model is difficult because there is a lot of human intelligence that also goes into this AI model. Besides, the data isn’t updated in real-time. Some information from 2021 might have changed now but that wouldn’t be reflected in ChatGPT. If something groundbreaking happens recently then that wouldn’t have a place in ChatGPT.
Also, ChatGPT was overhyped just like Alexa. Today Alexa is a loss-making division. Dall-E has a specific use case while Bloomberg GPT will also have a specific use case. In this way, moving forward AIs with specific use cases will have more utility than a generic CahtGPT. Lastly, OpenAI was supposed to be a non-profit open for all AI research companies but today it is owned by one of the big companies Microsoft. Hope you found this video valuable, please let me know which other topics you’d like me to cover in the future.
About Me
I publish meaningful and valuable content on this channel. My aim is to make business news more accessible and easy to grasp. If you find my videos informative and insightful then make sure to subscribe and leave a comment. I’ll see you in the next video
*****
Chapters
0:00 - Intro
1:17 - Chapters
1:32 -What is ChatGPT?
6:19 - Bias
8:34 - Current relevance
10:08 - Nature of ChatGPT
10:51 - Overhyped
12:47 - Takeaways
14:26 - Outro