In my previous post, I talked about the effects of generative AI on the travel and hospitality industry. As mentioned in the post, there are important ethical questions related to biases and stereotypes that should be considered.
A lot of the data in Generative AI is currently from Western sources which can cause a lack of diversity of race, ethnicity, beliefs and so on. Systems that are trained iwith Western data, they will only be able to produce recommendations based on this data. It is possible that the system then discriminates against a group of people or even shares stereotypical recommendations based on assumptions that it has made (Dogru et al., 2023). In the travel and hospitality industry, this can become an issue when Tripnotes.ai would only recommend spa’s, shopping malls and nail salons to female customers. Such recommendations are unethical and reinforce stereotypes. Following from Tripnotes.ai’s privacy policy, their recommendations are influenced by the data that they have collected about you. There is a chance that these recommendations are then influenced by bias and stereotypes. This can have severe negative impacts on the use of generative AI in the travel and hospitality industry.
To minimize the risk and possibility of such bias and stereotypes, Marinucci, Mazzuca and Gangemi (2022) suggest setting up a so called ‘’wordnet’’ which is an online tool that allows the addition of several sources of data to link relationships between used terms. Such a tool can help to weaken the relation between women and spa’s, de-biasing algorithms as it goes. It also creates more transparency for users, that can see what relationships are assumed.
As the ethical questions are addressed and we are aware of the bias, generative AI can be incredibly helpful in the travel and hospitality industry. It can make better, more personalized and less costly travel recommendations. For my next trip, I am looking forward to using Tripnotes.ai for some inspiration, are you?
Sources:
Dogru, T., Line, N., Mody, M., Hanks, L., Abbott, J. A., Acikgoz, F., Assaf, A., Bakir, S., Berbekova, A., Bilgihan, A., Dalton, A., Erkmen, E., Geronasso, M., Gomez, D., Graves, S., Iskender, A., Ivanov, S., Kizildag, M., Lee, M., … & Zhang, T. (2023). Generative artificial intelligence in the hospitality and tourism industry: Developing a framework for future research. Journal of Hospitality & Tourism Research, 10963480231188663.
Marinucci, L., Mazzuca, C., & Gangemi, A. (2023). Exposing implicit biases and stereotypes in human and artificial intelligence: state of the art and challenges with a focus on gender. AI & SOCIETY, 38(2), 747-761.
Tripnotes.ai. (2019) Privacy Policy [Terms and Conditions]. Consulted from https://welco.me/privacy.
I find your post very interesting, but it encouraged me to think why companies would actually engage in training its models on different, unbiased data. From a social equality and progressive perspective it of course is indeed a must, however, for companies that in principal are trying to beat others in the market, often by driving their operational costs down, why would they engage in that? Who should encourage them?
I think to address this issue, one of the possible solutions in to increase social awareness and by that increase pressure on companies to be less biased. Additionally, I believe there should be policies in place that analyse the data used by firms and oblige them to adhere to greater inclusivity. Lastly, it is worth highlighting that by offering a more diverse set of options to customers, not only the ‘biased’ results, firm can reach new customers.