How ChatGPT helped me learn Portuguese

9

October

2024

No ratings yet.

I remember vividly the moment I found out about ChatGPT. We sat in a tutorial and were asked to read an article in 5 minutes and discuss the content with our group. And there it was: “I’ll just copy-paste the text here then we will have a summary”, I heard from two chairs away.

I was not convinced, what was my teammate talking about? Was this safe? When I saw the result, I was shocked, that summary was simply perfect, just what we needed to complete the task. That day, as I went home, I couldn’t help but wonder: in how many ways could this chat make my life different? It didn’t take long to find out. A few nights later at girls’ night, my friend who had just broken up with a guy, was asking ChatGPT how to reply to a message to best convey her emotions. A few months later, I came back home for the holidays, and my mom was asking ChatGPT for a recipe to cook with a few ingredients that were about to expire. My mind was blown. Was this new technology going to change forever the way we approach tasks both in an educational context and in our personal lives?

Fast forward one year. I was in Lisbon for my exchange, and although it was not mandatory as the university was completely in English, I really wanted to learn Portuguese. Since I could not enroll in additional courses, and an external course was too expensive, I decided to try and learn it on my own. Being in a university with only international students, it was very hard to spot a native speaker. Where was I going to find someone who was available for free, accessible from any place I was in, who would not judge me, and who could keep up with my night owl schedule? That’s when I had the idea of asking ChatGPT to become my Portuguese conversational friend.

After studying some basics of Portuguese on my own, I was finally ready to experiment. I have started by asking simple questions, such as showing me a list of common words or sentences used in everyday life.

After realizing the potential of ChatGPT in this field, I have started trying to type longer sentences, asking if it could correct my sentences and give me tips to improve.

Soon enough we were having conversations, at first very simple ones, and each time I would increase the level of details of the conversations, for instance, writing prompts such as “Imagine we are at the sports club..” and giving more context to enrich the quality and increase the level of the conversation. That generated quite fun scenarios where the chat became really accurate about the characters and the role it was playing in specific contexts.

Sometimes it would generate some awkward statements, like when duo lingo makes you learn sentences such as “the cat is drinking water”, however as I used ChatGPT more and more, the tool itself also learned how to better respond to my prompts based on my needs.

Overall, these are just some ways that I used to get acquainted with a new language, however, there are many other creative ways to explore the potential of ChatGPT, for instance by creating engaging exercises to improve your grammar and vocabulary while doing something fun, making the whole process of learning more attractive[1].

As I reflect on it now, it is amazing to think how many uses we can make of GenAI. My experience with ChatGPT in the process of learning Portuguese was extremely positive, and even though my level of Portuguese is not the best, I was still able to make significant improvements. While before I was always a bit skeptical and would only think of ChatGPT as something that was taking over our way of reasoning and creativity, for the first time I was able to look at this technology as something extremely useful, that if used with moderation and acknowledging its limits, could be a helper, rather than a substitute for my abilities.

Would you try ChatGPT to learn a language? What are other creative and useful ways you were able to exploit this technology?


[1] Ter, S. (2024, June 19). Learning English with ChatGPT: A Comprehensive Guide. Medium. https://erkanyasun.medium.com/learning-english-with-chatgpt-a-comprehensive-guide-538b9b55832c

Please rate this

AI-generated Deep Fakes and US elections: A new threat to electoral integrity?

25

September

2024

No ratings yet.

“Wait, did I see correctly? Is Taylor Swift really supporting Donald Trump in the upcoming 2024 US presidential elections? That does not seem right”
If you woke up to this news on your social media and felt confused, don’t worry, you have just fallen into the trap of deepfakes, and you are definitely not alone.

It wasn’t long ago that Cambridge Analytica took the personal data of millions of Facebook users without their consent to use them for political advertising purposes, influencing the result of the 2016 US presidential elections[1]. Today, technology is again being turned against voters to manipulate them and affect electoral results. With AI-generated deepfakes, people can create compromising videos, images, or audio of anything and anyone, including persons of influence, to make them do or say anything they decide, and you will not always be able to distinguish between what is real and what isn’t. The Taylor Swift case was not isolated. Earlier this summer a deepfake of Democratic nominee Kamala Harris’s voice making provocative statements on delicate political issues was released reaching over 150 million views[2]. Similarly, fake AI-generated pictures of Donald Trump with Jeffrey Epstein have become viral[3]. Many other examples are out there in which either candidate is being targeted, or false election dates and general information are altered. With the widespread use of Generative AI, virtually everyone, including voters, can create deepfakes, spreading false information to affect other people’s decisions and challenging electoral integrity.

You would think that you as well as other average people would be immune to this and that you have the tools to detect when content is fake. Well, it is not that easy, as the AI technology behind the generation of deepfakes is being refined every day, making the content extremely accurate. CBS Boston’s I-Team made an experiment showing people a deepfake video in which one of their reporters was making a speech that was actually given by the Massachusetts Governor. The results of the polls were shocking, with more than half the people surveyed not being able to recognize that the video was fake.  

While different states are regulating the use of AI-generated fakes differently, Federal laws restricting the use of deepfake, especially for elections, are still in the making. In Texas, for instance, the use of deepfakes is considered a criminal offense, however, in other states, regulations are either less strict or non-existent. In the meantime, as voters, we have to remember to always check and recheck facts and be skeptical and critical about what we see and spread on social media, and the potential results of a misuse of AI technology. Ironically, AI can also come into help when trying to detect deepfakes.

As the first AI-disinformation-fed elections in history are approaching, the threat of deepfakes needs to be addressed, especially in the light of future elections, educating voters to spot what is real and what isn’t, but also through unified and strict regulations condemning the generation of this content.

What is your opinion on the subject? Do you think deepfakes will really influence the results of the upcoming elections? Do you think earlier actions in developing Federal regulations could have mitigated this phenomenon?


[1] Graham-Harrison, E., & Cadwalladr, C. (2021, September 29). Revealed: 50 million Facebook profiles harvested for Cambridge Analytica in major data breach. The Guardian. https://www.theguardian.com/news/2018/mar/17/cambridge-analytica-facebook-influence-us-election

[2] Tenbarge, K. (2024, August 2). Bipartisan effort targets deceptive political AI deepfakes [Video]. NBC News. https://www.nbcnews.com/tech/misinformation/kamala-harris-deepfake-shared-musk-sparks-free-speech-debate-rcna164119

[3] Sayegh, E. (2024, May 28). The Battle for Truth in election seasons: AI-Generated Deepfakes. Forbes. https://www.forbes.com/sites/emilsayegh/2024/05/14/the-battle-for-truth-in-election-seasons-ai-generated-deepfakes/

Please rate this