Is it morally and ethically appropriate to use ChatGPT (or similar AI system) for work and education-related written communications?
With both a pre-teen and a full-fledged teenager at home attempting to use ChatGPT for homework assignments (contrary to my advice regarding its negative impact on their ability to problem solve and think critically), I have come to question whether I should be using ChatGPT myself for work-related communications. It seemed a bit hypocritical. If I ask my children to think and write on their own, shouldn’t I?
In May 2023 Emma Goldberg wrote an article on this subject for the New York Times, claiming that we spend up to 5 hours a day writing emails – a disproportionately large part of a workday. Shortening and making this time more effective, and even creating better and more to-the-point responses feels like a no-brainer.
Emma experimented by using ChatGPT to write all her work communications for one week.
The most prominent result? The AI bot, unsurprisingly, could not capture the emotional tone of her standard online conversations. One co-worker even answered “OK, wait, let me get the ChatGPT to make a sentence.”
Capturing emotional nuance appears to be the most difficult challenge of AI. Despite their increasing sophistication and interactive capabilities, AI bots cannot escape “programmed” lineage. Designed to assist humans in specific mechanistic tasks, they do not yet have the capability to make emotional connections or create the proper or intended tone of a conversation.
Despite their limits, can (and should) AI bots assist people with disabilities who might be struggling to express themselves? There are clearly benefits.
Through the application of advanced natural language processing capabilities, people with disabilities can find assistance to write and express their thought process more accurately and comprehensively. For people on the autism spectrum, AI can also help bridge the communication gap they may experience.
In Kazakhstan, one interesting project aims to provide the blind and visually impaired better access to information in both educational and professional settings. In the study, researchers introduced the integration of cloud-based automatic speech recognition (ASR) and text-to-speech (TTS) technologies with ChatGPT, to provide individuals with visual impairments a natural interface for information access.
In this application of AI, the system architecture comprises ChatGPT, a messenger bot (Telegram), ASR, and a TTS module. The messenger bot acts as a user interface, enabling users to engage with ChatGPT. Visually impaired and blind individuals submit their requests to ChatGPT through voice messages via the messenger bot. Subsequently, the ASR module transcribes the voice messages into text, as ChatGPT processes text-based prompts. The ChatGPT module then analyzes the text inputs and generates responses in textual form. These textual responses are transformed into audible speech through the TTS module. Ultimately, the text and voice responses are delivered to the messenger bot for user interaction. This innovative configuration works in both English and Kazak now.
AI is rapidly penetrating our daily lives – and will become more integral to work and education in the future. As this trend continues, decision makers must be diligent to understand how, and the extent to which, people are applying it, and both the positive and negative implications of such use.
References
https://www.nytimes.com/2023/05/07/business/ai-chatbot-messaging-work.htmlhttps://www.techrxiv.org/articles/preprint/ChatGPT_for_Visually_Impaired_and_Blind/22047080