prova_1

Teaching and Learning Innovation Hub

Practical Guide for Students, Faculty and Staff

Luiss AI Handbook

Luiss

What should you be aware of?

What is AI?

What AI can do for you?

Luiss AI Policy and General Rules

What would you like to know?

Learn more

Learn more

Learn more

Learn more

Artificial Intelligence (AI) is a scientific technique that enables machines to mimic intelligent human behavior. Artificial General Intelligence (AGI) refers to the ability of a machine to perform any task a human being is capable of. Current capabilities can be positioned along what is called the “jagged technological frontier”. These capabilities of producing human-like work have improved rapidly, especially since the release of Generative AI, a category of Machine Learning technology that is able to produce new multimedia content based on existing data patterns. Examples of such GenAI tools include text-based chatbots like ChatGPT, reaching 100 million users within just two months of its release in 2022, breaking all-time records of speed in user adoption. AI and Generative AI are changing the way people learn, work and live – there are great advantages but also limitations

What is AI?

source

Compute refers to the computational resources and processing power required to train, develop and run AI models. These resources include central processing units (CPUs), graphics processing units (GPUs) and specialized hardware like tensor processing units (TPUs).

Data is the feedstock of AI. Feed the AI model text-image pairs and the AI develops the ability to recognize or generate images. Feed the AI model music and it will compose music. "You are what you eat" perfectly captures the importance of data as an AI building block.

Models are sophisticated computer programs designed to perform tasks that would typically require human intelligence. They are built on foundations of algorithms and large datasets that enable them to analyze patterns, make predictions and solve complex problems

AI basics

source

GenAI

DL

ML

AI

AI basics

As the capabilities of AI are rapidly evolving and poorly understood, it can be hard for professionals to grasp exactly what the boundary of this frontier might be at a given point. To get an intuition of where is this frontier, it is important to distinguish what is meant for “intelligence”, and its unique features when associated with “artificial intelligence” compared to “human intelligence”.

Task outside the frontier

Task inside the frontier

Task inside the frontier

AI abilities Equal difficulty tasks ---

Source: Dell’Acqua et al., 2023

AI capabilities cover an expanding, but uneven, set of knowledge work that has been represented by the "jagged technological frontier”. This is a “virtual” line, moving through time as technology advances, and separating tasks “inside the frontier” from tasks “outside the frontier”. People who skillfully navigate this frontier gain large productivity benefits when working with the AI, while AI can actually decrease performance when used for work outside of the frontier.

Capabilities and theory of mind

• AI is rapidly devising algorithms that “think humanly,” “think rationally,” “act humanly,” and “act rationally” (Csaszar and Steinberger, 2022); as shown also in the words used to introduce the new model Strawberry and chain-of-though mechanism that GPT-1o proposes.

• AI from its birth as science aimed to model machines as human minds. However, from its premise it took a theory of the mind as computation: AI builds on the idea that cognition—both by machines and humans—is a generalized form of information processing, a type of “input-output” device. This is not the only view that exists to analyse human mind and intelligence. There should be awareness of others point of views that highlight specific strengths of human intelligence and artificial intelligence.

Source: Felin and Holweg, 2024

Capabilities and theory of mind

Generative AI benefits

Learn more

Key differences to understand

Prompting

Generative AI uses

Two discoveries have made the history of what is called Generative AI: the transformer and the related attention mechanism. Generative AI tools are input-outputs machines. The input is given by the user and with the new Multimodal AI can be in different formats (text, image, video, audio, code). The output is different depending on the specific tool, and can be text, image, video, audio, code as well.

Generative AI

Prompting

Key differences to understand

Take a look at the LLM arena

What is an API? In the context of Generative AI, an API (Application Programming Interface) is a set of rules and protocols that allows different software applications to communicate and interact with each other. It acts as a bridge between the generative AI model and other applications, providing a standardized way for them interact and to... How to stay updated on the latest Generative AI models available and on their performance?

Key differences to understand

Manufacturing

Software development

Financial services

Advertising & Marketing

Media & Entertainment

Healthcare

source

Generative AI is impacting many industries. Take a look at some applications across:

Generative AI uses

Learn about its limitations

A Productivity Booster Generative AI has the potential to significantly boost global productivity. A McKinsey report (2024) estimates that it could add $2.6 trillion to $4.4 trillion annually to the global economy. This represents a 15-40% increase in the impact of all AI. Impact on the workforce:

• Increased labor productivity: Can boost productivity by 0.1-0.6% annually.

• Worker transitions: Requires investment in training and skill development.

Benefits

Staff member

Faculty member

Student

I am a …

What can AI do for you?

GenAI can help you rephrase text to improve tone, grammar, length or style of communication; change the wording suggesting synonyms. Text-based chatbots like Copilot, ChatGPT, Gemini can help you adapt your content to whatever tone or style of communication; Grammarly to check the grammar and conciseness/clarity of your text while typing; ; DeepL Write, that improves the quality of the text changing tone and style. DeepL online translator; Canva (premium) can translate automatically content in a document in whatever format (infographics, A4, presentation etc.) Notebook LM (premium) to help you summarize and organize research; Consensus to get an overview of research pieces related to a specific research question Text-based chat-bots like Copilot, ChatGPT, Gemini can generate questions or provide feedback to your preparation on a topic; Poe.com (premium) can be used to generate a chat-bot fed with your course content with which you can interact and ask questions; Notebook LM can produce audio summaries and targeted questions to test your understanding of the topic.

AI can be used as:

• Writing assistant:
• Translator
• Research assistant for assignments
• Exam or Class preparation tutor

AI x Students

AI x Students

Chatize or poe.com can be used to chat and summarize main points from a doc in whatever format Canva (premium) can be used to create content in many different formats (docs, presentations, social media formats), it can be used for image or video generation starting from text inputs or to brainstorm ideas, it can create AI avatars that animate your contents and turn them into audios or videos, in different languages (useful for this purpose also MyGen for videos up to 3 mins); Wix (costs related to publishing your website) can be used for automated website generation; Notebook LM for podcasts generation; some tools offer credits to generate images for free, like Leonardo.ai, Microsoft Bing Image creator, or for videos ⁠Runway ⁠Gen3, veed.io, ⁠⁠Luma Dream Machine. LearnWise (premium) can give you access to 24/7 student support powered by AI, in the form of a chat-bot dedicated to answer whatever question related to your university/courses Vmock (premium) can offer AI-powered resume and interview guidance; it can evaluate and give feedback on your cv or your performance in job interviews.

AI can be used as:

• Summarize documents or get quick answers on specific sections/aspects
• Creation assistant of multimedia content
• Personal support 24/7
• Enhance your employability and/or assess your competencies

Teaching AI can be used as a:

• Syllabus creation support- text-based chat-bots like Copilot, ChatGPT, Gemini can give you ideas on how to structure the syllabus; openAI offers a specific bot for AI curriculm integration, which gives you some ideas on how and where (i.e., for which activities or in what moments) to integrate AI in your educational CV.
• Discussion and debates facilitator – with Poe.com (premium) you can try role plays defining a character students can talk with; text-based chatbots like Copilot, ChatGPT, Gemini can generate suggestions of questions for in-class discussions or exams; can write drafts for case studies that align with specific learning objectives; can write templates or drafts for students’ feedback; can be helpful sources for assignment personalization and examples, rubrics for assessments…
• Students’ tutor 24/7 – Poe.com or Socrat.ai (premium) can be used to create a chat-bot based on course content (that learns from it) to provide personal tutors to students;
• Content creation support in different formats – Canva (premium) is also useful to create content in multiple media and size formats, brainstorming with its AI; veed.io or MyGen are able to produce text-to-speech videos with or without avatars, in multiple languages. Notebook LM can be used to transform text into an entire podcast, simulating a conversation of one or more speakers on a certain input. With such tools you can easily animate your teaching material or make it more interactive and engaging for students.

AI x Faculty members

AI can be used as a:

• Writing assistant - Generative AI chat-bots (Chat GPT; Copilot or Gemini) can help you formulate a first draft of a document or guide you on just the structure of it. You can ask them to rephrase your content, to avoid grammar errors, increase conciseness or change the tone of communication (e.g., making it more formal or friendly.) AI-enabled solutions can directly assist you while writing something and are embedded in your doc editing app (e.g., Grammarly provides live suggestions and proofreads your content, signalling errors or improvements, and changing them in just one click directly in the text); DeepL Write, that improves the quality of the text changing tone and style.
• Retriever of information and summarizer - Chatize or poe.com can be used to chat and summarize main points from a doc in whatever format
• Instant translator - Many tools can provide instant translations in different languages of your content/docs; Deepl, up to a certain amount of text data, can recognize the language used in some portion of text (copied and pasted it on its website) and then provide the translation of it in the language you need; Canva in its premium version can directly translate text in docs (e.g., in presentations, with Canva premium, you are able to create instantly the same doc in whatever language you want)
• Content creation support in multimedia formats - Canva (premium) is also useful to create content in multiple media and size formats, brainstorming with its AI; veed.io or MyGen are able to produce text-to-speech videos with or without avatars, in multiple languages. Notebook can be used to transform text into an entire podcast, simulating a conversation of one or more speakers on a certain input.
• Project management support - Excalidraw turns text into graphs and timelines to assist you in representing stages of projects or some sort of process or grid; Asana AI can be used to track and monitor projects involving multiple phases and people; Microsoft Copilot integration in Office (premium) improves productivity by automating routine tasks, providing data insights and producing documents automatically.

Administrative tasks

Research AI can be used as a research assistant: Notebook LM (premium) can help in summarizing and organizing existing research and literatures; Consensus helps you get an overview of research pieces related to a specific research question you enter as input; Dovetail (premium) or Atlas.ti (premium) can support and streamline qualitative data collection and analysis.

AI x Faculty members

Generative AI chat-bots (Chat GPT; Copilot or Gemini) can help you formulate a first draft of a document or guide you on just the structure of it. You can ask them to rephrase your content, to avoid grammar errors, increase conciseness or change the tone of communication (e.g., making it more formal or friendly.) AI-enabled solutions can directly assist you while writing something and are embedded in your doc editing app (e.g., Grammarly provides live suggestions and proofreads your content, signalling errors or improvements, and changing them in just one click directly in the text); ; DeepL Write, that improves the quality of the text changing tone and style. Chatize or poe.com can be used to chat and summarize main points from a doc in whatever format Many tools can provide instant translations in different languages of your content/docs; Deepl, up to a certain amount of text data, can recognize the language used in some portion of text (copied and pasted it on its website) and then provide the translation of it in the language you need; Canva in its premium version can directly translate text in docs (e.g., in presentations, with Canva premium, you are able to create instantly the same doc in whatever language you want) Canva (premium) is also useful to create content in multiple media and size formats, brainstorming with its AI; veed.io or MyGen are able to produce text-to-speech videos with or without avatars, in multiple languages. Notebook LM can be used to transform text into an entire podcast, simulating a conversation of one or more speakers on a certain input. Excalidraw turns text into graphs and timelines to assist you in representing stages of projects or some sort of process or grid; Asana AI can be used to track and monitor projects involving multiple phases and people; Microsoft Copilot integration in Office (premium) improves productivity by automating routine tasks, providing data insights and producing documents automatically.

AI can be used as a:

• Writing assistant

• Retriever of information and summarizer
• Instant translator
• Content creation support in multimedia formats
• Project management support

AI x Staff members

Depending on the specific office you are in:

• LearnWise (premium) can give you access to 24/7 student support powered by AI , on several topics (you determine the knowledge base; data in input)
• Vmock (premium) can offer AI-powered resume and interview guidance;
• Poe.com (premium) enables you to create your own chatbot to advice students on extra-curricular activities to follow

AI x Staff members

Luiss AI Policy and General Rules Special rules

Your precious role

Key issues in academia

Limitations of AI

What you should be aware of?

Lack of Originality

AI in unable to innovate and produce original ideas that go beyond the input data. Critical and creative thinking remains a uniquely human ability.

De-skilling

Over-reliance on AI may produce a de-skilling of human competencies and capabilities., resulting in reduction of personal critical thinking and ethical judgment.

Intellectual Property

Using AI tools in research or content creation raise concerns about intellectual property , unintentional plagiarism and academic integrity.

Opaque Decision-Making

The so-called “Black Box Problem”. i.e., not understanding how the AI’s internal processes work, how inputs and data are used to produce specific outcomes.

Privacy and Data Security

Personal and sensitive data can be used as input data for AI systems, increasing transparency and data breaches risks.

Algorithmic Bias

AI is set on data that may contain biases resulting in the final outcomes, perpetuating and amplifying the original biases.

Limitations of AI

• Advice on assessment design: AI may support the redesign of assessment, in order to limit the possibility of students relying solely on AI tools.

• Interactive Learning: AI tools can foster interactive learning environments, like simulations and virtual reality, which can improve both comprehension, information retention and engagement.

• Innovate teaching and evaluations: AI can be used to tailor classroom activities, support discussion and customize students’ learning and experience. Moreover, it can function as a push toward active learning strategies and assessments that require a holistic approach to course materials that challenge acritical AI use

• Over-monitoring: it can become very time-consuming and stressful for instructors to effectively limit or prevent use of AI by students

• Misuse of AI: AI may be used for purposes beyond their intended educational scope, leading to unethical approach to AI tools.

• Hallucinations, biases and prejudices: basing on the inputs data provided, GenAI feedback can result as unreliable and provide plausible but incorrect information.

OPPORTUNITIES

• Competitive graduate students: introducing AI into student’s learning, will prepare them for their future and professional lives where AI will be integrated.

RISKS

• Academic integrity: risk of plagiarism or counterfeit assessments, essays and written exams.

Key issues in academia

Become Literate Develop a AI literacy that is not just technical knowledge about AI, but is multidisciplinary, including Ethics and Media literacy, and involves also skills and attitudes

Human uniqueness Practice to enhance your unique human skills like critical thinking, communication, collaboration and emotional intelligence.

Become future-ready You need to be prepared to a future work and personal life where AI is more and more integrated

Agency and responsibility You need to direct the technology towards your self-enhancement, personal flourishing and goals. You are in control of your choices and actions with AI

Your precious roleYou have to consider yourself at the centre of this technological revolution

Curiosity and Continuous Learning

Promotion of exploration and learning about the potentials and limitations of AI, in order to use it effectively and responsibly in the educational context.

Openness in seeking assistance in case of difficulties or concerns regarding AI use, reaching out to the coordinator or faculty member. Simultaneously, committing to report any inappropriate behaviors or abuses in the use of AI tools, contributing to a safe and respectful learning environment.

Commitment to using AI tools honestly and respectfully, avoiding plagiarism and correctly attributing credit to the original authors of the information.

Support and Reporting

Sensitivity to Bias and Prejudices

Ethical Use

Verification of Information

Critical evaluation of AI-generated outputs, comparing them with reliable and verifiable sources to ensure the accuracy of the information.

Commitment to preserving one's online security and privacy, avoiding the disclosure of personal information and adopting cybersecurity practices.

Recognition of the limitations and potential inaccuracies in AI outputs, adopting a critical approach in interpreting and utilizing the generated information

Awareness of the benefits and risks associated with AI tools, committing to their responsible and ethical use, considering the implications of one's actions.

Privacy and Security Protection

Responsibility in AI Use

Awareness of Limitations

Luiss AI Policy and General Rules

I am a …

Staff member

Faculty member

Student

Specific Rules

Data protection

Protect Luiss account

Cite your sources

Privacy matters

Avoid entering confidential data (names, addresses, IDs) into AI tools.

Be accountable

Don't use your Luiss account for unauthorized AI apps or websites.

Acknowledge and reference any AI tools you use. See provided resources for proper citation styles. Example: MLA Style Guide or Monash University Guide

Ensure the AI tool's privacy policy aligns with Luiss University policies.

You are responsible for the content you produce. Check references, double-check facts, and avoid plagiarism.

Specific Rules for Students

Example 04 “[Tool Name] has been used to generate ideas and source articles. I was instructed to as part of this assessment. I have attributed the ideas it generated and my further development of them remains my own. I have read all source articles cited in this work."

Example 03 “[Tool Name] was used to generate alternative streams of thought. It was used to generate ideas to help me start my assessment. I have attributed the ideas it generated and my further development of them remains my own.”

Example 02 “[Tool Name] has assisted in structuring this assessment. This was used to help me organize my thoughts; the underlying work remains my own.”

Example 01 “I acknowledge the use of [insert AI system(s) and link] to [specific use of generative artificial intelligence]. The prompts used include [list of prompts]. The output from these prompts was used to [explain use].”

Responsibility in AI use Unless explicitly prohibited by your examiner in the assessment instruction, the use of specific tools or types of AI as sources for examples, texts, images, videos or audio files may be permitted as long as they are appropriately acknowledged and referenced. Specify what technology, if any, you have used to generate the materials you relied on for your assignment by adding a declaration that provides a written acknowledgement of the use of AI/Generative AI, including explicit descriptions of how the information was generated, the prompts used and explaining how the output was used in your work.

Specific Rules for Students

AI in Class: A Balanced Approach

Active Role in Grading

Insert critical component into assignments with AI so that students can assess how well or not its influence on students own thinking

Transparency

As you clarify which uses are acceptable and which are not by students, provide them information about how you used it (e.g., used the AI curriculum integrator chatbot to get ideas on where to integrate AI in your course activities).

Do not automate assessment

Using AI to brainstorm ideas for rubrics or further develop your initial idea of rubrics is fine; but fully automating grading is not acceptable. You should always play an active role in grading students and you should have the final say as you are responsible for assessments.

Beware of AI detectors

There is evidence they provide false positive; also they convey a negative perception and attitude towards whatever AI use. You should be able to discriminate acceptable vs not acceptable use.

Clear Guidelines

Insert the specific rules for AI use in your course in the syllabus document and be sure to take some time to present them to students and build your class awareness around them. Be very precise on which use is acceptable and which is forbidden. Clarify how you assess its usage.

Specific Rules for Faculty members

Turn some written assignments into other formats like pitches or oral presentations/exams to challenge them. Design tasks where students can demonstrate their knowledge but also transversal skills, as critical thinking or autonomous judgement.

Consider other formats

Promote class assessment activities, moving from the “output” to the “process” and enhance the importance of active participation in class as a significant element of the evaluation process and final grade.

Process-Oriented Assessment

Align with the university orientation on AI in courses: it is strongly encouraged to integrate AI in your CV, although not mandatory. The position you adopt should be transparent to students and should consider the side effects of banning policies (e.g., gender differences)

Use AI tools in class

Seek feedback from students on the effectiveness and fairness of AI use and integration in your course

Feedback

Specific Rules for Faculty membersTips for Responsible AI Use in Class

Experiences of other instructors on GenAI in class

T4L seminars

An opportunity to raise the bar!

Framework for AI integration in students’ tasks

PAIR framework

Syllabus policies’ collection for Generative AI use

Specific Resources for Faculty members

Data protection

Protect Luiss account

Privacy matters

You are responsible for the content you produce and hand in, hence, make sure it is trustworthy checking references included and double checking facts; make sure you are not plagiarising or using copyright-protected content improperly.

Specific Rules for Staff members

“We become what we behold. We shape our tools and then our tools shape us”Marshall McLuhan

Luiss

Deep Learning (DL) is an advanced subset of ML that leverages neural networks to learn from vast amounts of data. Neural nets function like the human brain, enabling systems to learn hidden patterns from data by themselves and build more efficient decision rules.

Machine Learning (ML) is a subset of AI that uses algorithms to automatically learn insights and recognize patterns from data, applying it to make better decisions. ML can be supervised (i.e., using labeled datasets) or unsupervised (i.e., finding patterns within unlabeled datasets). Popular applications of ML include computer vision and natural language processing (NLP).

The rationale is to distinguish what is within this moving frontier, i.e., tasks in which AI can complement or even outperform human work, from what is outside of the frontier, where AI output is inaccurate, less useful, and degrades human performance.

Human intelligence:

1. masters causal reasoning, i.e., the use of logic and facts to determine cause and effect relationships. As a detective investigating a case, it doesn't just gather facts or data, but tries to understand why things happened, building theories and connecting causes and effects.
2. is capable of disruptive creativity and future projection. It isn't just about reacting to stimuli; it is capable of “outside the box” or forward-looking thinking. It can imagine future scenarios, devise innovative solutions, and create something entirely and genuinely new. Consider the example of the Wright brothers who invented the first plane ever. Before their discovery, only birds could fly and only this “outside the box” thinking has led them to believe human flight could be possible.
3. understands cultural and social values, allowing to assess different situations more deeply and adapt strategies. It can grasp the nuances of language, identify emotions, and respond effectively.
4. masters developing long-term strategies. It considers multiple factors, assesses risks, and makes decisions that take a broader picture into account, and tacit knowledge.

Artificial intelligence:

1. masters quantitative information processing. As a statistical and computational model, it excels at crunching numbers in seconds and identifying patterns in past data. This makes it invaluable for tasks like data analysis, forecasting, and making quantitative data-driven decisions in stable environments.
2. learns from patterns in data. It's backward-looking, it is great at identifying correlations and making predictions based on what has happened before. AI doesn't understand the underlying causes of these patterns, but it makes predictions based on statistical analysis.
3. can integrate information and data from multiple sources with infinite levels of detail, organizing or summarizing them to create a more systematic understanding. AI can integrate data from various sources, such as social media, customer reviews, and sales data, to provide a complete and detailed view of a product's performance.
4. excels at automating repetitive tasks., performing them with speed, accuracy, and consistency. A customer service chat-bot can handle routine inquiries, as answering FAQs, by processing past information and recognizing patterns; or AI can analyze large amounts of data to provide recommendations.

• Exchange Data between the generative AI model and other apps. This data can include inputs for the model (e.g., text prompts, images) and outputs generated by the model (e.g., text, images, code).
• Expose Functionalities of the generative AI model, allowing other apps to access and use its capabilities. This includes tasks such as generating text, translating languages, writing different kinds of creative content, and answering your questions in an informative way.
• Integrate generative AI models into various apps and platforms. This enables developers to leverage the power of AI to enhance their products and services.
• Reach Scalability: APIs can handle large volumes of requests, making them suitable for applications with high usage. This scalability is essential for generative AI models that need to process and generate content efficiently.

The future of generative AI:

• Early pilots show promise
• Full realization of benefits will take time
• Leaders must address challenges

Overall, generative AI presents a significant opportunity for economic growth and improved productivity, but it requires careful planning and implementation.