The day ChatGPT outstripped its limitations for Me

20

October

2023

No ratings yet.

We all know ChatGPT since the whole technological frenzy that happened in 2022. This computer program was developed by OpenAI using GPT-3.5 (Generative Pre-trained Transformer) architecture. This program was trained using huge dataset and allows to create human-like text based on the prompts it receives (OpenAI, n.d.). Many have emphasized the power and the disruptive potential such emerging technology has whether it be in human enhancement by supporting market research and insights or legal document drafting and analysis for example which increases the efficiency of humans (OpenAI, n.d.).

Hype cycle for Emerging Technologies retrieved from Gartner.

However, despite its widespread adoption and the potential generative AI has, there are still many limits to it that prevent us from using it to its full potential. Examples are hallucinating facts or a high dependence on prompt quality (Alkaissi & McFarlane, 2023; Smulders, 2023). The latter issue links to the main topic of this blog post.

I have asked in the past to ChatGPT, “can you create diagrams for me?”  and this was ChatGPT’s response:

I have been using ChatGPT for all sorts of problems since its widespread adoption in 2022 and have had many different chats but always tried to have similar topics in the same chat, thinking “Maybe it needs to remember, maybe it needs to understand the whole topic for my questions to have a proper answer”. One day, I needed help with a project for work in understanding how to create a certain type of diagram since I was really lost. ChatGPT helped me understand but I still wanted concrete answers, I wanted to see the diagram with my own two eyes to make sure I knew what I needed to do. After many exchanges, I would try again and ask ChatGPT to show me, but nothing.

One day came the answer, I provided ChatGPT with all the information I had and asked again; “can you create a diagram with this information”. That is when, to my surprise, ChatGPT started creating an SQL interface, representing, one by one, each part of the diagram, with the link between them and in the end an explanation of what it did, a part of the diagram can be shown below (for work confidentiality issues, the diagram is anonymized).

It was a success for me, I made ChatGPT do the impossible, something ChatGPT said itself it could not provide for me. That day, ChatGPT outstripped its limitations for me. This is how I realized the importance of prompt quality.

This blog post shows the importance of educating the broader public and managers about technological literacy in the age of Industry 4.0 and how with the right knowledge and skills, generative AI can be used to its full potential to enhance human skills.

Have you ever managed to make ChatGPT do something it said it couldn’t with the right prompt? Comment down below.

References:

Alkaissi, H., & McFarlane, S. I. (2023). Artificial hallucinations in ChatGPT: implications in scientific writing. Cureus, 15(2).

Smulders, S. (2023, March 29). 15 rules for crafting effective GPT Chat prompts. Expandi. https://expandi.io/blog/chat-gpt-rules/

Please rate this

Sentiment analysis: How a computer knows what you’re feeling

5

October

2021

No ratings yet.

One of the most distinguished features of humanity, is being able to read someone else’s mood. People estimate other peoples mood and emotional state based on their verbal and non-verbal communication. Humans learn this skill from a very early age on and are able to distinguish the signs even for different people. For example, we all know when our best friend is feeling down, even when they’re trying to hide it for others. Maybe by being louder than usual, or they might be more quiet. But you know that they are troubled. Being able to distinguish this for different people is due to experience and how close you are to the other person. The years of friendship make it easier to read her mood and know when something is wrong.

Sentiment analysis

But what if I tell you a computer can also do this? And they don’t have to ‘be your friend’ for years and years. Using various techniques such as Natural Language Processing (NLP), the computer is able to recognise words and sentences as emotions. Words like ‘stress’ and ‘feeling alone’ are registered as negative, whilst ‘glad’ and ‘exited’ as marked as positive. This is called a sentiment analysis. Some sentiment analysis even link certain combinations of words to feelings, such as depressed, sad, cheerful or hopeful.

Possibilities

But what are the possible use cases of the sentiment analysis? And what are the challenges? Sentiment analysis is for example already being used in healthcare. Based on the unstructured notes a psychologist takes during sessions with their client, the sentiment analysis can track the clients mood over time. In this case, it includes possible diagnoses and ‘trigger words’ such as suicidal. This allows the psychologist to have one overview of the client’s emotional and mental state over time.

But what if this is taken one step further. What if this is done based on social media. Think about it: Your stories, chat messages, posts, web usage, emoticon usage, everything is documented and examined on your emotional state.

Challenges

One of the most important questions that arises from this, is regarding to privacy. Who is allowed when to measure your mental state based on your social media usage and your google searches? This is very sensitive (health)data, not stated by a doctor, but estimated by a computer. Another challenge is interpreting the words correctly, for example: ‘good’ is a positive word but ‘really not good’ is a negative combination of words. The complexity of languages can make it hard for a computer to interpret it correctly. However, Artificial Intelligence (AI) is often used for this so that the estimations become more accurate with each analysis.

Sources

Abualigah, L., Alfar, H., Shehab, M., & Hussein, A. (2019). Sentiment Analysis in Healthcare: A brief Review. In M. Abd Elaziz, M. Al-quaness, A. Ewees, & A. Dahou, Recent Advances in NLP: The Case of Arabic Language (pp. 129-141). Switzerland: Springer, Cham. doi:10.1007/978-3-030-34614-0_7

Denecke, K., & Deng, Y. (2015, March 25). Sentiment analysis in medical settings: New opportunities and challenges. Artificial Intelligence in Medicine, pp. 17-27. doi:10.1016/j.artmed.2015.03.006

Please rate this

Will IBM’s Watson Lead The Future Of Public And Political Debates?

2

October

2020

5/5 (2) Image source.

Weak Discourse & Polarisation

The first U.S. presidential debate of 2020 is yet again a reminder that there is a decline in civilised and substantive discourse. This becomes especially evident when personal attacks and interruptions are deemed the key moments of a presidential debate rather than the actual answers given to the questions [1]. Not to mention that some of the answers that were given often included misinformation and/or weak arguments. Bickering, squabbling, and providing unfounded arguments greatly diminish the quality of a discourse. It should have no place in any debate (for that matter) and definitely not in one between the presidential candidates of a country.

An increasing lack of wanting to really listen to others and their views in general is leading to even more political polarisation. While (social media) algorithms are filtering and personalising content which insulates people within a so-called filter bubble, it is the people themselves that are turning these bubbles into echo chambers through their own information behaviour [2]. These chambers (i.e. clusters) consist of users with similar interests or ideologies that are on a topic specific content or media ‘diet’. This diet often stems from the selective exposure of information proposition which entails that people prefer to be exposed to content that agrees with their own opinions. Only information that reinforces existing beliefs and conforms to (chamber) group norms is sought out and shared, even if this information is incorrect or substantiated by weak arguments [2].

With this type of behaviour on both social media and in political and public discourse, it becomes imperative to enable people to start listening to each other again and bring quality arguments to the table.

A Breath of Fresh Air

“That’s Debatable”, a new interactive debate show presented by Bloomberg in partnership with Intelligence Squared, could potentially be the breath of fresh air we so desperately need. What makes this show so unique is the use of IBM Watson’s advanced capabilities in Natural Language Processing (NPL) [3]. The show aims to present how AI can help people better understand nuanced viewpoints, inform decision-making in high-stake cases, and facilitate the discovery and surfacing of critical insights. Furthermore, it will enable a more diverse and larger range of opinions and voices to be brought to the public square as well as the providing new perspectives which can enhance a debater’s arguments [3].

So How Does It Work?

The general public is invited to submit a short statement in which they make a case for or against a specific topic [4]. IBM’s Watson then performs a Key Point Analysis, a new NPL capability, consisting of four phases to generate a coherent narrative. Firstly, a deep neural network classifies the submitted arguments into two distinct groups; supporting or contesting. Those arguments that are deemed off-topic, neutral, or irrelevant are removed from the pool. In the second phase, potential key points are identified by filtering and grading the quality of the arguments. Options that are too long, incoherent, include redundancies, or that have a tone that is too emotional are disregarded. Next, Watson assesses the prevalence of each potential key point by identifying the number of sentences articulating the gist of the points. The points are graded and ranked and ultimately a small set, that is diverse enough and covers the majority of the arguments, is selected. Lastly, Watson matches the strongest arguments to the specific key point it will support which are then used for the formulation of a coherent narrative that addresses both the pro and con side of the debate. This narrative will ultimately assist the debaters during the show to ensure a more well-rounded discourse [4].

Image source.

Image source.

Final Thoughts

With the rise in polarisation, I believe it to be incredibly important to advocate for healthy and substantive public and political discourse. “That’s Debatable” is definitely a step in the right direction and many countries would benefit greatly from a show like this being introduced in their own native tongue. However, there is a reasonable chance that the majority of the viewers will be individuals that are already more open to hearing what the other has to say. I wonder how much of a lasting social change can be achieved as long as people are not actively encouraged to remove themselves from their echo chambers and change their own information behaviour.

The show will premiere on October 9th, 2020 with the first debate focusing on the question whether it is time for wealth to be redistributed. If you want to be a part of the next discussion, you can visit IBM’s website. You have 15 days left (as of October 2nd, 2020) to contribute to the next topic: “A U.S.-China Space Race Is Good for Humanity” [4].

Sources

[1] CNN Politics (2020). Watch the key moments from the first presidential debate. [Online Video] Available at https://edition.cnn.com/videos/politics/2020/09/30/first-presidential-debate-highlights-orig-mh.cnn [Accessed: 2 October 2020]

[2] Zimmer, F., Scheibe, K., and Stock, W. (2019). ‘Echo Chambers and Filter Bubbles of Fake News in Social Media. Man-made or produced by algorithms?’, 8th Annual Arts, Humanities, Social Sciences & Education Conference, Prince Waikiki Hotel, 3-5 January. Honolulu: 2019 Hawaii University International Conferences

[3] Bloomberg Media (2020). Bloomberg Media and Intelligence Squared U.S. present a new limited debate series “That’s Debatable” hosted by Award-Winning Journalist and Moderator John Donvan and featuring AI from IBM Watson. [Online] Available at https://www.bloombergmedia.com/press/bloomberg-media-and-intelligence-squared-u-s-present-a-new-limited-debate-series-thats-debatable-hosted-by-award-winning-journalist-and-moderator-john-donvan-and-featuring/ [Accessed: 30 September 2020]

[4] IBM (n.d.). Join the debate | Experience a new era of improved public discourse with AI technology. [Online] Available at https://www.research.ibm.com/artificial-intelligence/project-debater/thats-debatable [Accessed: 30 September 2020]

Please rate this