OpenAI’s ChatGPT has created a furore in the IT community because of its accurate responses to professional questions. Let’s see how you can use this chatbot, also in combination with other software products.
I got interested in the technology to apply it in some of my work tasks as a UX designer. I would be happy if you shared your experiences too.
OpenAI, an artificial intelligence lab, created ChatGPT. It was launched a few years ago with financial support from Elon Musk and other entrepreneurs.
The bot is powered by the “big language model”, an artificial intelligence software that has been taught to predict the next word in a sentence based on analyzing vast amounts of internet text and finding patterns through trial and error.
Since its launch, the free version of the chatbot has been hugely popular. It received more than 1 million users in its first week. And this is even though access to it is restricted in several countries.
How compelling is ChatGPT?
Source: ADOBE STOCK
ChatGPT is so convincing it can easily fool people. It mimics text messages quickly, can create cocktail recipes, and write lyrics for songs and movie scripts. The chatbot convincingly answers additional questions logically related to the previous ones. This is seriously worrying for social media users. After all, how can one ensure that the lyrics are created by ordinary people and not by a chatbot?
Because of this, several bans on ChatGPT have followed. For example, Stack Overflow, a Q&A site for IT professionals, has banned ChatGPT texts. This was because the bot generated so many responses that they were “damaging” to the site.
Even OpenAI’s CEO tweeted last month that it was “a mistake to rely on it for anything important right now”:
Likewise, ChatGPT has been banned from NYC public schools. Teachers can’t tell the difference between student and student texts and chatbot essays. They believe ChatGPT is detrimental to the learning process. It does not build critical thinking and problem-solving skills, which are essential for success in learning and throughout life.
Nevertheless, OpenAI’s new AI tool could be a potential new online learning tool. After all, the chatbot quickly answers complex questions and provides detailed answers immediately. By contrast, users searching for information on Google on their own have to scan pages of results in detail until they find what they are looking for. And Microsoft has already bet on chatbot in this matter. The company plans to launch a version of its Bing search engine in March 2023 using a chatbot to provide answers.
The fact is that ChatBot and similar systems are dangerous because they pose several ethical and social risks.
For example, ChatGPT can:
Increase the spread of misinformation;
create computer viruses and facilitate hacking;
disrupt existing institutions or services by writing a good motivational letter for a job, student essay or grant application
promote fraud and copyright infringement;
reduce the value of intellectual work;
violate privacy, as AI systems often contain unethically collected data from Internet users.
In what areas does technology perform best?
I proceeded based on what a chatbot does best.
It was answering questions. ChatGPT knows how to answer questions better than other chatbots, and gives detailed answers to almost any question. I suggest offering a chatbot for the surveys we provide users to explore their experience with a product or app. This would be an interesting UX Research case study.
It was giving answers to search queries, like Google. The plus side is that the bot generates reasonably accurate and complete solutions, whereas the search engine produces a list of resources.
Write dialogues, essays, stories, and scripts and compose songs to suit the style of a particular artist. For example, ChatGPT recently created a piece in the style of Nick Cave, though the musician did not like it.
Generate content for websites and social networks based on user requests. This will make the work of copywriters and content managers much more accessible.
Compose emails and generate meta tags. Rereading and possibly tweaking the message before sending it out is better. But overall, the chatbot does an excellent job.
Develop simple applications. A chatbot can give you a code sample which can be used or optimized for a particular application. The main thing is to set the task clearly and specify the programming language. Of course, the users can still be judged with significant reservations. But this is sometimes better than starting from scratch. At least, there will be ideas in what direction to go.
Translate texts.
It can filling in briefs or reports (from examples).
Of course, incorrect handling of requests can happen. Here’s one the notable example:
15 UX-specific Chat GPT prompts to get started in AI-assisted design
Here are some template questions you can use to get started in Chat GPT. They’re simplified so you can add more context and get better results.
1) List design and research methods to help [challenge].
2) What’s the best way to design a solution to this brief: [insert brief].
3) Write a brief for [challenge].
4) Brainstorm design ideas around [opportunity].
5) Create wireframes for an [industry] experience for [user].
6) Create an SVG icon for [your client].
7) Write code for a [your project].
8) Write survey questions about [context] for a [user].
9) Write user tasks for a usability test on a [your project].
10) Act as a UX designer. Here is my first question: “[insert question]”.
11) Create study notes on this: [insert text]
12) What are the main user issues with [well-known product]
13) Summarise the problems in these interviews: [insert transcript].
14) Synthesize this text into a new design approach: [insert text].
15) Uncover potential user biases in [context].
ChatGPT learning resources and cases for UX designers (and not only)
