Category: AI

  • Redefining Tech with Ethics: “Whoever writes the code dictates the rules.”

    Redefining Tech with Ethics: “Whoever writes the code dictates the rules.”

    At the inaugural FPS2025 summit, one of the most captivating conversations centered on the intersection of artificial intelligence (AI), privacy, and the ethical implications of technology.

    Meredith Whittaker, President of Signal and a vocal advocate for tech accountability, engaged in a thought-provoking discussion with Matteo Flora, the most expert on AI and digital reputation in Italy and probably one of the most prominent at global level. Their conversation, which took place in L’Aquila, Italy, highlighted critical issues in the tech industry today — from the dangers of surveillance capitalism to the ethical dilemmas of AI deployment

    I decided to dedicate this post to this interview, that I found incredibly important and yet not seen / heard enough.

    The Code Dictates the Rules: Who’s Behind AI?

    A central theme of the conversation was the idea that “whoever writes the code dictates the rules.” Whittaker emphasized the profound implications this has for our society.

    With the rise of AI, we are seeing unprecedented scales of power being concentrated in the hands of a few tech giants who not only control vast amounts of data but also influence societal norms and political landscapes.

    AI, she explained, is not magic but simply sophisticated code executed on powerful hardware platforms. However, these systems are often built on biased data, reflecting the interests of a small group of corporations with a profit-driven agenda. Whittaker raised concerns about the monopolization of AI development, highlighting how only a handful of companies own the infrastructure and the data that train these systems. This creates a dangerous asymmetry in power, as these companies dictate the narrative surrounding AI, often casting it as a superhuman force that society should revere.

    The Ethics of AI and Data Collection

    One of the most pressing issues discussed was surveillance capitalism—the business model that underpins much of the tech industry. Whittaker contrasted this model with the mission of Signal, which is to provide a secure, private communication platform that respects user privacy. She pointed out that the current tech landscape, where personal data is collected and monetized, is not designed with the user’s best interest in mind.

    Signal, as a nonprofit organization, faces the challenge of maintaining its privacy-first model in an industry that thrives on data exploitation. Running such a platform at scale costs millions of dollars annually, and Whittaker explained that if Signal were a for-profit company, it would inevitably face pressure from investors to compromise its privacy standards for the sake of profit. This is why Signal must remain independent of the profit-driven business models that dominate the tech industry.

    The Global Impact of Privacy

    Throughout the conversation, the global implications of tech and privacy were a recurring theme. Whittaker discussed the importance of cross-border communication and how platforms like Signal empower people worldwide to connect securely. Unlike nationalized versions of tech products, which would be restricted by jurisdictional boundaries, Signal’s universal design ensures that it works seamlessly for people across different countries and cultures.

    In addition to tech’s global reach, Whittaker also emphasized how AI can contribute to the homogenization of culture, creating a risk of losing linguistic diversity and cultural richness. She cautioned against the dangers of AI reflecting only the narrow perspectives of a few dominant cultures, particularly those in Silicon Valley, and warned that this could stifle creativity and unique ways of thinking.

    Basically the message is: “Most – all – digital large global platforms are designed to collect user data, making it nearly impossible to build similarly large, stable, and user-friendly platforms without relying on data-driven ROI. The very basic infrastructure is nearly impossible to build without huge investment”

    Building a Better Future with Tech

    Despite the challenges and ethical quandaries, Whittaker remains optimistic about the potential for change. She advocates for innovating the business models behind tech, suggesting that true innovation lies not just in new technologies but in rethinking how these technologies are funded and distributed. She sees a future where privacy and fairness are not afterthoughts but foundational principles.

    The conversation at FPS2025 ended with a call for a societal shift. Whittaker urged policymakers to adopt regulations that challenge the current business models and protect individual rights in the face of rapidly advancing technology.

    She also urged the European Parliament to embrace a more equitable tech ecosystem, one that does not succumb to the dominant narratives set by large corporations but instead fosters a digital environment that works for everyone.
    Especially now that the UE signed a letter with 44 European CEOs asking the EU to pause the AI Act for two years, it is the time to rethink the legislation and make it suitable for innovating while ensuring parity and equal rights.

    The Bottom Line: Tech for Good

    Whittaker’s reflections underscore the importance of ethical tech in the modern world. As AI continues to reshape our societies, the conversation about who controls the code and how that power is used has never been more crucial.

    Through platforms like Signal, Whittaker and her team are pushing back against the status quo, demonstrating that it is possible to build technology that prioritizes privacy and fairness over profit. The challenge, however, remains: will we as a society allow this vision to flourish, or will we continue to let surveillance capitalism dictate the rules?

    In the end, Whittaker’s message is clear: the future of tech must be inclusive, transparent, and accountable. It’s up to us — the users, the builders, and the policymakers — to decide how we want that future to look.

  • TikTok and the Dopamine Connection: Why It’s So Addictive

    TikTok and the Dopamine Connection: Why It’s So Addictive

    In just a few years, TikTok has become a cultural phenomenon, captivating millions of users worldwide with its short, engaging videos and hyper-personalized content. Whether you’re watching funny skits, dance challenges, or life hacks, the platform keeps you hooked with its seemingly endless stream of entertainment.

    Some numbers around TikTok

    TikTok has experienced remarkable growth and engagement since its launch. Here are some key numbers – Source: How many users on TikTok? Statistics & Facts (2025)

    • Content categories: The most popular categories on TikTok are Entertainment, Dance, and Pranks, with billions of views each
    • Monthly active users: TikTok has over 1 billion monthly active users globally, with 150 million in the United States (growth of over 10 times in last 5 years).
    • User demographics: About 44% of TikTok users are under 25 years old.
    • Time spent: The average daily time spent on TikTok has more than doubled from 27 minutes in 2019 to 58 minutes in 2024.
    • Engagement rate: TikTok has an average engagement rate of 5.96%, making it the most engaging social media platform.
    • Regional growth: By 2023, TikTok is projected to have 682 million users in the Asia Pacific region (excluding China and India), 238 million in Europe, and 192 million in North America.

    But what makes TikTok so hard to put down?

    The secret lies in its algorithm, which curates content tailored to your interests, and its ability to trigger the brain’s reward system through bursts of dopamine—the “feel-good” chemical that reinforces pleasurable activities. This clever combination of psychology and technology creates a feedback loop that keeps users scrolling for hours.

    Dopamine is at the heart of TikTok’s addictive design.

    The app capitalizes on variable rewards—those unpredictable moments when you find an exceptionally engaging video—similar to pulling the lever on a slot machine. Each time you laugh at a funny clip, discover something new, or feel validated by likes and comments, your brain releases dopamine, reinforcing the desire to keep scrolling.

    TikTok’s endless feed and short-form content further amplify this effect, delivering quick hits of entertainment without requiring much effort. The result?

    A platform that not only entertains but also conditions the brain to crave more, making it one of the most compelling social media experiences today.

    A 2022 study from the Frontiers in Psychology journal found that TikTok is the most highly addictive of all social media platforms. TikTok manipulates dopamine levels through several key mechanisms designed to create an addictive user experience.

    TikTok’s AI-driven algorithm

    TikTok’s AI-driven algorithm is a sophisticated system designed to create a highly personalized and engaging user experience.

    TikTok’s AI-driven algorithm is a highly sophisticated system that tailors content to individual users, creating an addictive and personalized experience. By analyzing user interactions—such as likes, shares, watch time, and even pauses—the algorithm learns preferences and predicts what content will keep users engaged.

    It uses machine learning to identify patterns in behavior and pairs this with metadata from videos, including hashtags, captions, and trending sounds, to deliver a feed uniquely curated for each user.

    This real-time adaptability ensures that TikTok’s content always feels fresh and relevant, driving high levels of user engagement and fostering an unparalleled connection between the platform and its audience.

    Why It Can Be Addictive

    1. Short-form content: TikTok’s 15-60 second videos provide rapid-fire bursts of entertainment, each potentially triggering a small dopamine release[1]. This creates a short-term dopamine feedback loop, reinforcing the behavior of continued scrolling. According to a 2020 study from the University of California – San Diego, TikTok users in the United States open the app an average of eight times per day and spend an average of 46 minutes on the app.

    2. Infinite scroll: The endless stream of content removes natural breaking points, keeping the dopamine system constantly engaged and making it difficult for users to stop scrolling. According to the researchers of another study “the system quality has a stronger influence than information quality in determining adolescents’ experience with TikTok and that the flow experience has significant direct and indirect effects on TikTok addiction behavior” – meaning the way the algorithm chose the content to visualize and for how long influence more the experience than the content itself.
    3. Personalized “For You” page: TikTok’s AI-driven algorithm presents highly tailored content, creating a consistently rewarding experience that triggers more frequent dopamine releases. The average daily usage of TikTok is 95 minutes (compared to 17 minutes for YouTube and 29 minutes for Instagram), made possible through continuously analyzing user behavior and content characteristics, TikTok’s AI algorithm creates an addictive and personalized experience that keeps users engaged for extended periods.
    4. Variable reward pattern: The algorithm regularly incorporates new content types to discover new facets of user interests. The unpredictable nature of content quality in the feed triggers dopamine release, similar to the mechanism behind slot machines. Besides dopamine, TikTok also triggers the release of epinephrine, norepinephrine, and oxytocin, creating a complex neurochemical response that enhances the addictive potential.
    5. Real-time adaptation: The algorithm quickly adjusts to changes in user preferences and behavior. Using a new sophisticated “collisionless embedding” system, it organizes not only content but also user data efficiently (by categories) enabling better and faster recommendations. The “Monolith” system (this is how it’s called) updates its model continuously, learning from user interactions and syncing the data at short intervals.
    6. Content relevance over creator popularity: Even new creators can gain visibility if their content resonates with users. It mixes constantly contents from new creators with established ones, to enhance the possibilities to become viral for anyone, also without high number of contents.

    This combination of features creates a powerful cycle of anticipation and reward, keeping users engaged and continuously seeking the next dopamine hit. The result is a “dopamine machine” that can lead to addiction-like behaviors and potentially negative long-term consequences for users’ mental health and cognitive function.

    US is concerned: TikTok faces a U.S. Ban

    The potential TikTok ban in the United States stems from national security concerns over its parent company, ByteDance, and its ties to China. Lawmakers worry that Chinese government laws could compel ByteDance to share sensitive user data or leverage TikTok’s content algorithms to spread misinformation. As a result, legislation has been passed requiring ByteDance to sell TikTok to a U.S.-approved buyer or face a nationwide ban. Despite TikTok’s denial of these allegations and efforts to distance itself from ByteDance, the pressure on the platform remains high.

  • Rethinking AI in Advertising: The Missed Opportunity in the Gemini Olympics Ad

    The recent Gemini ad that aired during the Paris Olympics has sparked significant discussion, and I find myself agreeing with some of the criticisms it has received. The ad, which portrays AI assisting a father in writing a letter for his daughter, left many with the impression that AI was replacing the father’s role in an intimate and personal task.

    In a recent interview with The Wall Street Journal, Google’s Android chief Rick Osterloh defended the company’s decision, stating, “Clearly the market isn’t fully ready to embrace all the changes that come with AI.” I believe this is a very unfortunate way to respond to the criticism.

    Shifting the responsibility to the audience, implying they are “not ready” or didn’t understand the message, misses the point entirely and is never a good idea. It’s not about the market being ready—it’s about understanding the deeply personal nature of the relationship the ad depicted and how AI was framed in that context.

    Instead of taking responsibility for the misstep, this response seems to place the blame on the audience, which only further alienates those who felt uncomfortable with the ad’s message.

    Rick further insisted that AI technology is now so powerful that it would allow Sydney McLaughlin’s daughter to write several appreciation emails or even produce different kinds of content, like an essay or a poem. But again, this misses the essence of the ad’s intended message. The purpose of this communication campaign was to enhance the emotional bond between father and daughter in the context of sports, ideally with the thoughtful integration of AI – not to promote the “text writing” capability of Gemini, as though it were the only “understandable” and easy-to-grasp feature for the audience.

    Moreover, it’s important to consider the broader implications of the ad’s portrayal. The fact that only Black people are represented in the ad could be misinterpreted as suggesting that they are “not educated” and need help with writing because they are not capable themselves. This kind of messaging is not only misguided but potentially harmful.

    Instead of focusing on AI’s ability to write text, the campaign could have shown how AI can genuinely support and strengthen relationships, particularly in the context of a father-daughter bond in sports. For instance:

    • Performance Analytics: AI could have been depicted helping Sydney McLaughlin analyze her race performance, offering insights to refine her technique and improve her times.
    • Personalized Training Plans: The ad could have shown AI assisting her coach in creating a personalized training regimen tailored to her unique strengths, weaknesses, and goals.
    • Mental Preparation: AI could also play a role in mental conditioning, offering meditative guides or visualization exercises to help Sydney maintain her focus under pressure.
    • Injury Prevention and Recovery: An AI could have been portrayed helping her monitor her health and recovery, ensuring she stays in peak condition throughout her training.

    These examples would have demonstrated AI’s capability to augment human connections and enrich shared experiences, rather than replacing personal moments or perpetuating stereotypes.

    As we continue to explore the boundaries of AI, it’s crucial to strike the right balance and ensure that we’re enhancing human connections, not diminishing them. And when feedback comes, let’s be open to it—acknowledging the need for better messaging and deeper understanding, rather than attributing it to a market not being ready.

  • ChatGPT-4o and how communication is never going to be the same again

    ChatGPT-4o and how communication is never going to be the same again


    It’s about time I start talking about ai and directly sharing my experience with ChatGPT-4o. I have just put my hands on it today and realize this is going to change everything forever in my profession as well as in the business in general.

    Today I want to share one small use case that I happened to test for my work that I thought would be interesting for more people to read. Hope you enjoy it!

    Exploring Consumer Awareness and Acceptance of Alternative Proteins with ChatGPT-4o

    As someone deeply involved in the food ingredients sector, I’ve come to realize that understanding the relationship between end consumers perception and communication of source ingredients of their food is becoming key.

    This connection is the link between consumer satisfaction and experience with the finished product and the information and messaging (claims) on food packaging/promotions. It’s a complex interplay, as every piece of information must be backed by rigorous scientific studies and regulatory approvals but it should also speak and evoke something meaningful for the consumers.

    This is why nutritional brands and food product producers/manufactures have always been relying on science and published researchers for their nutritional statements and in general the product communication around “health” (in the broader sense).

    Historically, accessing and analyzing research was a laborious task. It required a deep knowledge of where to look, long hours of sifting through journals and publications, and, before the advent of digitalization, even physical visits to libraries. The digital age brought some relief, making scientific papers more accessible online. However, finding the right keywords, navigating through Google Scholar and other platforms, and keeping up with the latest research still demanded significant effort.

    All of that is now over with ChatGPT-4o and specifically Scholar GPT: this is a revolutionary tool that is set to change the landscape of research and communication in our industry. This advanced AI not only simplifies

    • process of finding relevant studies
    • integrating data from online public articles and academic papers
    • providing a comprehensive view of the topic at hand
    • even evaluating papers in various languages and formats.

    To illustrate the power of ChatGPT-4o, let’s test the GPT with a real case. I used a topic of increasing importance as example and wanted to analyze what literature is available on consumer awareness and acceptance of alternative proteins.

    With just a simple 3 steps approach – I asked the AI to help me find studies on this subject and summarize the key findings.

    Image of process of creation of alternative protein
    created with DALL-E

    Here’s an example of how ChatGPT-4o transformed my research process:

    1. Just prompt the research topic in ScholarGPT: “Findings on Consumer Acceptance of Alternative Proteins”

    Here is the first result:

    1. A Systematic Review on Consumer Acceptance of Alternative Proteins
      • Findings: Plant-based meat substitutes are the most accepted among consumers, followed by cultured meat. Insects are the least accepted. Consumers are more willing to try alternatives that closely mimic traditional meat.
      • Source: ScienceDirect
    2. Meta-Analysis of Consumer Acceptance of Alternative Protein Sources
      • Findings: Analysis of 91 studies indicates that consumers are generally open to alternative proteins but have varying levels of acceptance. Familiarity and perceived naturalness positively influence acceptance.
      • Source: Faunalytics
    3. Stakeholder Beliefs about Alternative Proteins
      • Findings: Stakeholders, including consumers, express both enthusiasm and skepticism about alternative proteins. Health benefits and environmental sustainability are significant drivers, while concerns include taste and price.
      • Source: MDPI
    4. Consumer Insights & Research Reports on Alternative Proteins
      • Findings: Detailed insights into consumer demographics show that younger, health-conscious individuals are more likely to adopt alternative proteins. Motivations include health benefits, ethical concerns, and environmental impact.
      • Source: Good Food Institute
    5. Alternative Proteins: Consumers’ Perceptions Analyzed
      • Findings: Key drivers for acceptance include taste, nutritional benefits, and environmental impact. Barriers include unfamiliarity and skepticism about health claims.
      • Source: ETprotein
    6. 2020 Consumer Research Priorities
      • Findings: Identifies key research priorities to better understand consumer attitudes, including taste preferences, nutritional knowledge, and environmental awareness.
      • Source: Good Food Institute PDF
    7. The Untapped Climate Opportunity in Alternative Proteins
      • Findings: Highlights the growing consumer acceptance of alternative proteins and their significant potential to reduce environmental impact. Consumers are increasingly motivated by climate concerns.
      • Source: BCG
    8. Alternative Protein Consumption: A Systematic Review
      • Findings: Consumers show a positive trend towards alternative protein consumption, with plant-based proteins leading in acceptance. Cultural factors and awareness significantly influence consumption patterns.
      • Source: Wiley Online Library
    9. How Are Different Alternative Proteins Perceived by Consumers?
      • Findings: Plant-based proteins are perceived more favorably compared to insect-based and cultured meats. Factors such as taste, nutritional content, and ethical considerations play a crucial role in acceptance.
      • Source: Kerry Health and Nutrition Institute
    10. How Well Are Alternative Protein Sources Received by End Consumers?
      • Findings: Alternative protein sources are generally well-received, with increasing consumer preference for products that are sustainable and have a minimal environmental impact. Taste and texture improvements are essential for wider acceptance.
      • Source: Phys.org

    2. Prompt: “Write a Summary of researches outcome” but do that with ChatGPT

    If you use ScholarGPT to write the summary the result is not good to be honest. I tried to simply continue working within Scholar but the summary it gave was not based
    on previous researches but they were taking into consideration other sources, much more promotional, for example:

    1. Overall Acceptance Levels:
      • Plant-Based Proteins: These are the most accepted alternative proteins among consumers. Familiarity with plant-based foods and their ability to closely mimic traditional meat are significant factors driving this acceptance​ (The Independent)​​ (BarBend)​.
      • Cultured Meat: Cultured meat holds moderate acceptance. While consumers are intrigued by its potential environmental and ethical benefits, concerns about naturalness and taste remain barriers​ (The Independent)​​ (Verywell Fit)​.
      • Insects: Insects are the least accepted alternative protein. Despite their nutritional benefits and low environmental impact, cultural aversions and unfamiliarity hinder widespread acceptance​ (The Independent)​​ (BarBend)​.

    As you can see, the considerations were correct but the sources were probably based on the popularity of the websites rather than on their scientific relevance.

    I actually had to copy the text from Scholar and include it in a new ChatGPT Prompt as directly from ScholarGPT the first outcome was not accurate.


    Still the summary “remembers” also the results from Scholar and the results were much better – see the final outcome.

    Final outcome:

    Alternative proteins, including plant-based, cultured, and insect proteins, are emerging as sustainable solutions to traditional animal-based proteins. Understanding consumer acceptance and awareness of these alternatives is crucial for their successful market integration.

    Research indicates that consumer awareness of alternative proteins is increasing due to enhanced media coverage and marketing efforts, though understanding of specific types like cultured meat and insect-based products varies.

    • Consumers are motivated by the perceived health benefits and lower environmental impact of these proteins, significantly influencing their willingness to try and adopt them.
    • Sensory attributes such as taste, texture, and appearance are critical determinants of acceptance, with products closely mimicking traditional animal proteins being more readily accepted.
    • Cultural norms and social influences also play substantial roles, with higher acceptance in regions where alternative proteins are traditionally consumed.
    • Effective marketing strategies and educational initiatives, emphasizing transparency about production processes and benefits, can further enhance consumer acceptance by building trust and interest.

    Image of alternative created with DALL-E

    3. Save with a name convention!


    All responses in ChatGPT 4o are still saved in the same sideboard without distinction from which GPT is used. As you can see it is impossible to know which prompt was used in which GPT and if you try to massively use it, the list of prompts on the side is going to get huge!

    It’s important to establish a clear naming convention to differentiate prompts based on the specific GPT utilized. This practice ensures clarity and helps in efficiently retrieving the correct information. MM stands for my own GPT (Maria Morganti) which I’ll explain later.


    My Takeaways:

    1. Diverse Document Processing: The research included PDF documents that had been scanned and correctly interpreted by ChatGPT-4o, showcasing its ability to handle various formats seamlessly.
    2. Language Translation and Inclusion: Impressively, one of the sources was in German. ChatGPT-4o translated, evaluated, and included this source in the summary in real-time, demonstrating its multilingual capabilities.
    3. Accuracy of Summaries: The summaries of each finding were accurate, reflecting the information from the original sources without distortion.
    4. Limitations in Summary Generation: When I prompted ChatGPT-4o to create a summary akin to a seminar paper, the result was disappointing. The AI generated new text based on other, non-scientific sources that were mainly promotional and lost the connection to the original research.
    5. Comparison with Standard ChatGPT: In contrast, when I used the standard ChatGPT for the same task, the text was much more neutral and accurate, maintaining fidelity to the initial research findings.


    By leveraging ChatGPT-4o, I was able to quickly gather and synthesize relevant research, providing a clear and concise overview of consumer attitudes towards alternative proteins. This not only saves time but also ensures that the information shared is accurate and backed by credible sources.


    Make your own GPT

    To facilitate this process in the future, I made my own Researcher GPT that is already set-up to do this work, in the way I need to (with the right sources) and in the tonality I want to. I can’t wait to further test it my “Research Summarizer” GPT…See below a brief introduction to my research tool:

    As we continue to navigate the evolving food landscape, tools like ChatGPT-4o will be indispensable in bridging the gap between scientific research and consumer communication, ultimately fostering a more informed and receptive market. Stay in touch for more updates from my digital Journey!