On the importance of AI literacy
Artificial Intelligence (AI) is rapidly changing the world around us. It is already being used – and it has been used for several years – in a variety of domains and on a variety of tasks, from education to medical diagnosis, from personalized recommendations to content creation. Still, in recent months, there has been a huge increase in the interest towards AI across a much wider population with respect to previous years, and this has been mostly due to the huge improvements in generative-AI and the public release of tools such as ChatGPT, Midjourney, and Stable Diffusion. This increase is clearly visible in the Google Trends below, which show how the number of AI-related searches drastically increased in the last 12 months.
Search term: Artificial intelligence
Search term: Generative AI
Search term: GPT
Search term: Text to image AI
The trends shown above clearly show that there has been an increased interest in AI – especially generative AI – and this is likely to continue in the future. Most importantly, even in the case that such interest across a variety of different demographics decreases in the future, the role of AI and AI-generated content in our society will not: students will be using GPT-like language models, and AI-generated images will be used by content creators to quickly prototype several ideas, just to name two applications of generative AI which are already common. Similarly, there will be malicious users of AI as well: AI-generated images and videos will be used to spread fake news – displaying something that never happened, such as the fake photos of Trump’s arrest that went viral earlier this year – and AI-generated news website could quickly be created to be used as sources to support fake news. This raises major concerns about media reliability, as it means that it will be always harder to understand whether an image posted online is real or artificially generated. There are already automated systems in place (and being developed) to detect such uses of AI, but this will be an arms race between the two sides, and it will not be possible to rely only on automated countermeasures. We will have to adapt to the existence of these new tools, and change how we interact (and trust) online content.
This change will not affect only fake news and the spread of misinformation, either. For instance, students will be using language models for answering exam questions, and educational assessment must adapt in order to be robust to these changes.
In this quickly evolving society, it is important to be AI-literate, and be aware of what are the tools and technologies that are out there, what are their capabilities, and how they work. This clip from a TED interview is a good example of why it is important to be AI-literate. In this blog post, we will go through some of the steps that you can take to become more AI-literate.
Understand the basics of AI
Having to deal with AI and AI-generated content, and having to understand whether the media that we are looking at is legitimate or artificially generated, it is important to have an intuition of how AI models work. It is not necessary to know the details of each algorithm, but rather know how these models are trained, and what is their objective.
Most generative AI models could be referred to as imitative AI models, as they are trained using large quantities of data and try to mimic what they have seen at training time. For instance, when you ask a question to a large language model such as ChatGPT, it will answer one word at a time and, for each word, it will output the word that is most likely to follow the previous text (taking this probability from the training dataset).
Be aware of the tools and their capabilities
This is slightly different from the previous point, as this considers the specific tools that are developed and available out in the wild. It is unfeasible to know the names of all the tools and the details of each of them, but it is important to be aware that modern models are capable of producing text, images, videos, and audios that resemble human generated content, and we do not live anymore in a society where we can trust on the things we see with our eyes. Also, it is important to be aware of the tools and their capabilities in order to leverage them, and the best thing to do is to experiment and play with these AI tools. Indeed, AI literacy can empower individuals to use AI effectively, such as for personal productivity or to solve complex problems. People that are unaware of (or do not have access to) AI tools are likely to be left behind in some career paths.
Be aware of the limitations
AI tools are far from perfect, and they will always have (big) limitations. They are not always accurate, they might answer users’ questions with statements that are factually incorrect, and can exhibit bias and discriminative behaviors. AI literacy can help individuals to understand that the output of AI tools should not be “blindly trusted”, and to prevent (or at least report so that it will be mitigated) the reinforcement of existing biases. Also, being aware of the limitations of AI systems can be helpful to detect whether some content that you are looking at is artificially generated: for instance observing the hands in an image.
Be concerned about privacy and security
AI models are trained using large quantities of data and, in many cases, personal and sensitive information. As such, AI can post significant privacy and security risks. Targeted advertising campaigns are used to push consumers into buying more products, and there have already been cases of this data being used for targeted political advertisement (such as the Facebook–Cambridge Analytica data scandal. A best practice, in this sense, is to modify the privacy settings to opt out of trackers, and possibly use private online tools (search engines, emails, cloud storage, etc.).
Be aware of AI legislation
In many countries the legislation about the usage of AI tools is playing catch-up with the technology, but things are starting to change. It is important to be aware of these developments as they will have huge effects on the AI tools that are available and their capabilities. For instance, in the US a lawsuit accused AI content creators of misusing copyrighted works as AI-generators such as MIdjourney are trained on billions of copyrighted images, and ChatGPT was temporarily banned in Italy over privacy concerns.
In summary, AI literacy is crucial in today’s world because it enables individuals and organizations to understand the implications of AI on society, and make informed decisions (not only about AI-related technologies). It is a complex landscape, and it is constantly evolving, but it is important to be aware of the benefits, risks, and implications of modern AI systems, as they will keep getting more present in our daily lives.