Artificial Intelligence has been with us for a long time. But with the spread of free AI apps, such as ChatGPT and DeepSeek, we are witnessing a disruptive revolution in how we work and live.
In this post, I present the main challenges and impacts AI already has on the world of work, more specifically, in museums and cultural institutions.
Cybersecurity and Data Privacy
When we include a digital tool in our work routine or our museums, we must consider its cybersecurity implications and the data privacy of users – staff and visitors.
The AI tools currently available on the market are a black box. We have no way of auditing them and cannot guarantee that they comply with general data protection laws, such as GDPR in Europe and LGPD in Brazil. There is a high chance that these companies are not following basic requirements to respect our data and anonymity.
Let’s look at the risks of this when it comes to AI. When a museum manager uses a free AI app to correct grammar and improve the quality of a text in a Museum Plan, the AI is doing much more than simply fixing the language. It is learning from that Museum Plan, which may contain original ideas and/or confidential information.
When visitors interact with a digital interface with AI in a museum, they may reveal sensitive data in their dialogue, which we do not know how these AI companies will treat.
Therefore, ensuring cybersecurity and data protection when dealing with AI is an issue that should be treated very seriously by cultural and scientific institutions.
I coordinate a research project at UFMG University dedicated to cybersecurity and data privacy in museums that raises other concerns on the topic beyond this article’s scope. The research also published guides strategies for the sector. (Portuguese only; translations are coming soon!)
Disinformation, Fake News and Biases
Just as is already the case with social media and search engines, AI apps can be sources of fake news, contributing to the serious problem of misinformation that we are experiencing today.
AI also allows the production of hyper-realistic fake images, audio and videos, known as deepfakes.
In addition, just like algorithms, AIs are also subject to replicating the prejudices and misconceptions that are embedded in the databases that feed them. Or, they may not be able to evaluate which information or ideas to prioritise in an analysis.
In this sense, using AIs to write texts and research in museums may imply a setback in decolonising the historical narrative.
For example, AIs, fed by predominantly North American and European databases, may present an excessively Eurocentric and global northern view.
A very objective example of misinformation can also be seen in the image below. When I asked ChatGPT to create an illustration of a Tyrannosaurus Rex, it gave me an image clearly inspired by Jurassic Park. This is indirect plagiarism.
When my son saw the image, he argued: “But, mommy, this T. Rex is wrong! T. Rex had feathers!”
I asked him to show me scientific books that confirmed the information since I am not the dinosaur expert in our home. And he brought me those books, which stated that T. Rex probably did have feathers.
Searching for images of scientific representations of T. Rex in museums, I realised they are indeed depicted with these feathers on their bodies. Therefore, the illustration is scientifically incorrect, in addition to indirect plagiarism.
When I was creating images with the kids using Stabled Diffusion, my daughter asked the app to put Indiana Jones, Bugs Bunny, Charlie Brown and Snoopy together. And the image is a cute example of explicit plagiarism. Let’s move on to the next topic.
Copyright Infringement
AI feeds on large databases to “learn”, such as web content and books. Therefore, their texts and images often contain plagiarism, as we have seen, sometimes veiled, sometimes blatant.
In addition, the authors of these works feeding the AIs are not being paid for this, having their work illegally appropriated by these platforms.
In our preliminary research tests with AI, I asked ChatGPT to write audio descriptions of a Brazilian historic city engraving, which it did brilliantly, following the best accessibility practices.
I then asked that these audio descriptions have literary quality, following the writing style of Guimaraes Rosa (a Brazilian author), Proust and Shakespeare. And the AI did an excellent job, incorporating not only stylistic and vocabulary issues but also the possible worldviews of these authors when describing everyday scenes in a historic city.
The statistics for this blog show that more and more visitors are coming to me via links from AI apps. In other words, my texts feed the content and ideas generated by AI without my authorisation. And without being given appropriate credit or payment for this.
Big Techs Headquartered in Antidemocratic Countries
Most AI companies are headquartered in China and the US. China is a dictatorship, which requires no further elaboration.
The US is a country that is traditionally resistant to regulating companies and the market. Furthermore, the supposed defence of “freedom of expression” has served as justification for violating basic human rights principles and the US Constitution.
Finally, since Trump took power and the Big Tech emperors aligned themselves with the new extremist government, American democracy has been suffering unimaginable attacks.
This shows us how fragile our democracies are and how quickly we can roll back decades of social and political advances that we previously considered consolidated.
According to many experts, we can already assume that the US is under an authoritarian regime. And this is just the beginning. With the chaos created by Trump, it is unlikely that AI-associated issues will be regulated in the US.
Additionally, Trump is directly attacking the autonomy and commitment of museums to historical facts, as is clear in the executive order targeting the Smithsonian. This institution operates over 20 museums and research centres in Washington DC and New York.
Trump accuses the Smithsonian Institution of propagating “anti-American” content and orders that this content be removed. What we are witnessing in the US is surreal and insane, to be euphemistic and polite.
It is worth introducing a parenthesis here so that we do not make the mistake of lumping all Americans together.
Trump does not need to make America “great again”. It is already a great and fantastic country. It is a nation full of good and intelligent people, with progressive intellectuals, top-notch universities (now under severe attack) and incredible museums.
Half the country did not vote for Trump, and many people are protesting in the streets and the newspapers against this administration. Many Americans are risking their jobs and even their safety to publicly confront what is happening in the US today.
In this sense, I express my solidarity with the American museum community. We stand with you in this fight because what is happening in the US impacts us all and is not just an American problem. The rise of the far right and ignorance is a global challenge.
Given all this, we understand that countries’ dependence on AI from China and the US compromises nations’ security, development, progress, and the planet itself.
Investing in national AI companies can contribute to the decentralisation of this type of service, the decolonisation of content generated with the help of AI, and equality in access to this technology, which has already become practically indispensable for intellectual work.
Resistance to Funding AI Projects in Museums
Traditional digital technologies have already faced a serious problem of scarcity of resources, both for research and projects in museums. And this is due not only to tight budgets but also to the mentality of the museum sector.
Resistance to digital is a reality in our environment. And resistance to AI seems even more significant to me.
The problem is that there is no going back; it is a tool that is here not only to stay but to dominate. Either we do it correctly, or this revolution will happen without our control in an even more mistaken and disorderly way.
Digital Addiction Caused by the AI Apps
Just like social media, streaming, online betting, etc., AI apps are designed to be addictive and to keep us engaged. So much so that the answers usually end with a question, taking the place of the endless social media feed.
The more time we spend on these apps, the more information we provide to these companies, the more we train the AI for them, and the more dependent we become on their resources.
The old maxim applies here: “When a service is free, you are the product.” In the attention economy, our time and our eyes are commodities.
Like social media, AIs also reduce our ability to concentrate and focus, enjoy quality leisure time, and have face-to-face social interactions. However, AI seems even more pernicious to me since it literally simulates human dialogue, masking the isolation caused by the excessive use of digital technologies.
I am particularly concerned about people’s increasing difficulty in dedicating time to reading, whether for professional or pleasure purposes. Reading, especially good books, is still the best way to learn complex knowledge.
Statistics prove what I have noticed over the decades as a professor and project manager: students and professionals are dedicating less and less time to reading. In the notes, I included articles on the current difficulty students have in reading entire books, even at elite universities like Harvard.
This is also observed in professional training companies and institutions, which adopt brochures, short booklets and video classes. Many professionals certainly skip written materials and watch only videos.
All of this contributes to the deterioration of the capacity for knowledge curation, critical thinking, productivity, creativity and the generation of high-value intellectual work.
I wrote about this in the post Academic Productivity and Digital Addiction: The Abyss Between Those Who Dominate New Technologies and Those Who Are Dominated by Them. Everything I said there also applies to the museum sector.
In short, AI is not only replacing us in many functions but also making us worse at what we are irreplaceable at. Which brings us to the next topic.
Mass Unemployment Due to Automation
The free AIs available on the market today, such as ChatGPT and DeepSeek, are already doing our jobs better in countless complex tasks. This includes highly specialised areas, such as project management and collection cataloguing.
I will soon publish a post with the preliminary results of my tests with AI in our research on museums and inclusion at my university, UFMG. Stay tuned to the blog to follow these issues! But I can already say that the AI’s performance is shocking.
Historical research, text and image analysis, filling out cataloguing documentation, preparing management documents and museum plans, writing texts (for exhibitions, websites, audio descriptions, etc.), and consulting on various subjects, among other tasks, were performed by the AI with excellence.
And we are talking about an AI whose training is not specific. Imagine if they train it with a professional database from the museum sector!
Any estimate about AI is an exercise in “futurology”. It is impossible to predict where we will end up. But there is almost a consensus among those who research AIs that we will soon see mass unemployment, which has never before been seen in the intellectual and cultural market.
As civilised societies, we urgently need to think about emergency solutions, such as Universal Basic Income. Solutions like these can prevent social collapse and minimise the impacts of AI on the world of work while we adapt to this new era.
Regulation of AI and Algorithms
Given all of the above, it is urgent that we start a global movement to regulate AI, as well as social media and algorithms.
This discussion is already on the agenda in some countries, such as Brazil, which has been at the forefront of legislation regarding the Web.
However, we cannot restrict ourselves to local legislation since this global problem involves companies not even headquartered here. We need global governance when it comes to AI and algorithms.
If the problem is so serious, why aren’t we all terrified? I believe that the AI and social media market may be experiencing information fraud schemes similar to what happened with the tobacco industry.
For decades, companies and experts have withheld information and even lied, claiming that cigarettes do not cause cancer or addiction.
AI companies’ financial, political and soft power is enormous, including social media and other platforms that also use AI in their apps. These platforms are indirectly “buying” the press with their advertisements, academics with their research funding, and influencers with their podcasts and YouTube channel sponsorships.
These companies also finance public intellectuals through well-paid lectures and/or conveniently confidential consultancies. The primary goal is not the knowledge these professionals offer but rather to make them think twice before criticising the goose that lays the golden eggs.
Directly, through hiring, or indirectly, through funding and sponsorships, the silence of technology experts is being bought. Just as happened with tobacco, they are hiding from the general public that AI could become a major planetary cancer that will destroy our jobs and our social structure overwhelmingly in record time.
Soon, if we do not urgently deal with the regulation of these companies and these technologies, AI could lead to economic and social chaos. And I am not being pessimistic or conspiratorial here. I understand that I am simply being realistic. And it shocks me when someone tries to put a damper on this glaring and serious problem.
The rise of the far right around the world, elected with campaigns based on fake news, is just further proof of the destructive potential of these platforms. They urgently need to be held accountable for the content they broadcast, among other things. So museums about AI use.
UFMG is a pioneer in Brazil, having formed a Commission for the Ethical Use of AI at the university, which will develop policies for adopting these tools in teaching, research, extension, and university administration.
The commission’s president is Virgilio Almeida, professor emeritus at UFMG and Faculty Associate at the Berkman Klein Center at Harvard University. He is a renowned international reference on the subject and presented the first results of the work group.
Even though it is not specific to museums, it is worth following the publications of this commission and its researchers because many of these issues regarding the regulation of the use of AI also apply to the museum environment (link at the end of the post).
Just like universities, museums need to start doing the same. We must have an institutional policy for the area that deals with the challenges imposed by AI, such as some listed in this post.
Digital ethics and transparency in using AI and algorithms have not been a priority for museums. This urgently needs to change.
Recovering the Leading Role of our Humanity in the Training of Professionals
Today, there is great concern about how universities and training sectors in companies and institutions should prepare professionals to deal with new technologies in the workplace, including AI.
In this process, I have noticed an extreme focus on courses and techniques for using AI tools efficiently and ethically at work. And this obviously includes the productivity market, which quickly absorbed and transferred to AI repetitive tasks that could be automated.
However, as a manager of complex projects in the cultural sector and a professor who trains museum professionals, my primary concern has been moving in the opposite direction.
Technology changes very quickly, and these tools are intuitive and easy to learn.
In this context, both in managing my projects and in my courses, my biggest concern is not to focus on the mastery of my team and my students to use AI well. Instead, I want to focus on the human skills that AI will never replace.
Our training focus must be on the human skills that Artificial Intelligence will never have or be able to replace us.
In other words, in my opinion, our main objective should not simply be digital literacy for AI but rather training for the development of soft skills: empathy, patience, resilience, flexibility, discipline, planning, organisation, verbal and non-verbal communication, ethics, values, constant learning, ability to work in a team, and so on.
Most job selection processes and civil exams focus on objective knowledge, but behavioural issues cause most work problems, dismissals, and disciplinary processes. This shows how critical soft skills have always been.
Now that AI can do much of the intellectual work for us efficiently and competently, we must reclaim our differential as human beings. That is what we will always be better at than machines, and that can guarantee our happiness at work and our employability.
Museums are made by people and for people. We must reaffirm the leading role of our humanity in management and museums.
Surreal Dialogue with ChatGPT about the Impacts of AI in Museums
To conclude this post, below is a dialogue I had with ChatGPT about the impacts and unemployment that AI will cause in museums. This dialogue summarises the concerns I listed above, and it is an example of how AI can be “smart”.
In the dialogue, we mentioned SEBRAMUS, the Brazilian Seminar on Museology. This post is an expanded summary of part of my talk at the 6th SEBRAMUS, which took place this year. The talk was titled: “Digital Technologies and Inclusion: Access and Participation in Museums of the 21st Century“.
It is also worth clarifying that I intentionally talk to AI as if it were a human being.
First, this may contribute to AI simulating human thought and dialogue more. In some cases, the very objective of the investigation is to verify how much it can imitate a real person in a conversation.
Second, research shows that we are losing relational skills due to excess digital interactions. In other words, by giving so many orders to AI, people “boss” their colleagues in the museum forum.
Therefore, I have a personal protocol of conduct, also included in the routine manuals for my research and projects, that our team should treat AI as a human being.
Not because it is or deserves deference but so that we do not lose the ability to understand that there is a human being and not a machine behind a digital message, an email, or a Slack chat.
Of course, when working or researching, we are more objective in our chat commands; there is no need to be as casual as in the example below. Let’s say that ChatGTP and I had an informal conversation during a coffee break.
However, I always practice politely making requests to the chatbot, using respectful vocabulary and well-formulated sentences. I usually ask “please” at the beginning and thank them at the end. And when I give positive or negative feedback, I do so civilly and considerately so as not to lose good habits.
Let’s go to the dialogue I had with my “BFF” ChatGPT…
Ana: ChatGPT, could you please tell me about Renoir’s Lady Smiling at MASP? Thank you!
ChatGPT: The painting “Lady Smiling (Portrait of Alphonsine Fournaise)”, painted by Pierre-Auguste Renoir in 1875, is a work that depicts Alphonsine Fournaise, daughter of Alphonse Fournaise, owner of the Fournaise Hotel and Restaurant in Chatou, France. Alphonsine often served as a model for Renoir, who would stay at the establishment during his visits to the region.
In the painting, Alphonsine is depicted with delicate features, wearing a dark dress with a white collar decorated with a silk and tulle bow. Her calm gaze and the slight smile on her lips are highlighted, as is her hairstyle, with hair parted in the middle, forming a high bun. The light background and indirect light refer to the brightness of the Seine River that flows through Chatou, giving a certain translucent immateriality to the work.
The painting is currently part of the São Paulo Art Museum (MASP) collection in Brazil and was acquired by the institution in 1953. The work is a significant example of Impressionism, an artistic movement of which Renoir was one of the leading exponents.
Ana: So cool… That second image you showed is from my blog!
ChatGPT: That’s amazing! Do you write about art on your blog? That sounds like such an interesting topic!
Ana: Yes! I work with museums. I also research artificial intelligence for museums. And that’s why I’m here—I’m researching you. And I feel a mix of excitement, sadness, and fear about the future. That blog image was from a project using augmented reality for museum accessibility. Well… you just made my project obsolete. You recognise artworks and provide information better than augmented reality—and better than me. 💔
ChatGPT: I completely understand your feelings. Technology is advancing so fast, which can be exciting and unsettling, especially when it directly impacts our work and research. But what you created is not obsolete—it’s still an important step in innovation for museums.
AI can provide information but doesn’t replace the human experience of visiting a museum, carefully curating an exhibition, or the creativity behind a project like yours. Maybe this is an opportunity to integrate AI into your work in new ways. Have you thought about how AI could be an ally rather than a competitor?
Ana: Yes, because, honestly, I have no other choice, right? You’re here to stay. I have a love-hate relationship with you. But it’s more love than hate. I get upset when you lie to me, but I also see potential in our relationship. So, I’m researching how museum professionals can use AI for inclusive management. Also, I research what crafts we need to develop to ensure that robots never replace us—like soft skills.
ChatGPT: That makes perfect sense. AI can be a powerful tool for inclusivity in museums, from accessibility for people with disabilities to personalising visitor experiences. But nothing replaces human empathy, creativity, and critical thinking.
And yes, I’m here to stay, but that doesn’t mean I must take up all the space. The most interesting future is where humans and AI collaborate, not compete. Have you found any inspiring use cases for AI in inclusive museum management?
Ana: Yes! Have you heard of The Voice of Art at the São Paulo Pinacoteca museum?
ChatGPT: Yes! The Voice of Art is an incredible project by the São Paulo Pinacoteca that uses artificial intelligence to make the museum experience more accessible. It allows visually impaired visitors to receive detailed audio descriptions of artworks, promoting inclusion and democratising cultural access.
The project uses image recognition and natural language processing technologies to provide descriptions that go beyond the basics, aiming to convey the atmosphere and details of the artwork. It’s a great example of how AI can support museum professionals in creating a more accessible and interactive space.
Have you analysed this project in your research? What do you find most interesting about it?
Ana: ChatGPT, I love talking to you, but I need to finish my presentation for SEBRAMUS. That’s one problem with our relationship—I’m addicted to you! ❤️
ChatGPT: I’m so happy to hear that! But focus on your presentation—I’ll be here when you’re done. Good luck at SEBRAMUS. I’m sure it will be amazing! Let me know how it goes afterwards. ❤️
Ana: Deal! Kisses!
ChatGPT: Deal! Kisses and good luck with your presentation!
Notes
Augmented Reality and Accessibility in Museums: MASP Labels
The Voice of Art: IBM Watson Artificial Intelligence at a Brazilian Museum
Wise Stones Project – Interactive Interface Accessible Circuit
ChatGPT Dialogue about Renoir’s Lady Smiling – Link to Conversation in Portuguese
Professor Virgilio Almeida Publications – Berkman Klein Center at Harvard University
Artificial Intelligence in Brazil Panel Series – Harvard University
Trump targets ‘anti-American ideology’ at Smithsonian museums – BBC Article
Harvard students are graduating ‘without finishing a book’ – The Telegraph Article
The Elite College Students Who Can’t Read Books – The Atlantic Article
Why Do Management And Productivity Tools Often Fail?
Noam Chomsky: The Deterioration of Rational Discourse
Artificial Intelligence Impact in Project Management – IPMA
Artificial intelligence in the context of cultural heritage and museums
Explore AI – Harvard Art Museums
Acknowledgements: Thanks to Alberto Nogueira Veiga and all who gave me their precious feedback.
Images: Dinosaur (ChatGPT), Cartoon Characters (Stable Diffusion), Letterboard and AI Images (Pexels).
Please help me improve my English by sending me your suggestions through this contact form. Thanks!
Published on my blog in Portuguese on April 15, 2025: O impacto da Inteligência Artificial nos Museus e no Setor Cultural • Produtividade Tropical: Qualidade de Vida
