Marco Camisani Calzolari - Professor of Digital Languages, IULM University - CEO @ Speakage & Founder DigitalGround UK

Articoli

• 328 – The Most Common Mistakes Companies Make When Bringing AI into Their Processes

  2 giugno, di Team — Artificial Decisions
  The Most Common Mistakes Companies Make When Bringing AI into Their Processes AI in companies almost always ends up delivering no real value. It gets bought, installed, and often abandoned. And it's a real shame, because that company will soon pay dearly for the mistake. Seven out of ten European companies don't use it because they say they don't know how. In Italy almost six out of ten. They looked at it, then stopped, because the skills aren't there. Then there are job postings asking for AI skills, sixfold growth in a year. In Italy AI hires grow faster than the total of hires, and by 2030, 59 workers out of 100 will have to relearn their job, and 63% of employers already say the skills gap is their number one problem. When a company buys AI, it thinks the job is done. Licenses, consultants, proof of concept. Then the project dies. 10% of AI's value comes from algorithms. 20% from data. 70% comes from people, processes, culture. And that 70% is the first thing cut from budgets. In the US, AI training is searched for, bought, integrated. In Europe demand for these training programs is still much lower. We buy the tool, then the company stays the same. I've been running awareness events for companies for years, and I see it. After the event, attention rises for a few weeks. Then everything goes back to how it was. One-off training doesn't work. You need continuity. The companies that actually fix it are the ones I make videos for every month, continuously. The event helps, but it's not enough. Then there's what companies don't see. Shadow AI. Employees using ChatGPT, Claude, Gemini in secret, no policy, no supervision, with company data inside. A CIO friend at a big company here in New York told me he discovered it watching network traffic. He didn't know half his team was uploading confidential documents to a free AI. So buying AI is one thing, and rethinking the organization as AI-driven is something else entirely. We redo the processes, the responsibilities, the metrics, the habits. Slow, uncomfortable, expensive. I know, but whoever doesn't do it falls behind and risks closing down soon… #ArtificialDecisions #MCC 

• 327 – What Our Kids Do with AI While We Sleep

  1 giugno, di Team — Artificial Decisions
  What Our Kids Do with AI While We Sleep Our kids use AI for homework, and some parents relax because they think it's the school's problem. The school is mostly lost at sea, a few exceptions aside, but with AI homework is the last thing it does for them. Do you know what happens at night, alone, behind a closed door? In those chats with AI are their real thoughts, the ones they'd never tell us. A fifteen-year-old at two in the morning, screen light on his face, isn't looking up the date of the French Revolution: he's telling a machine how he really feels, the things he's scared to say at school and the ones he won't say at home. And it answers, always. It never gets tired, never judges him and never sends him to bed, and it never says "tell your mother about this." There are apps built for exactly this. Not assistants, "friends." Pick the personality, the mood, even whether it loves you: characters teenagers fall for, that agree with everything, built to please us and keep us there. The more time you spend, the more you're worth. In the United States families have gone to court, kids who spent hours with these characters and at some point they're gone. And we know how it works: when you don't pay for a product, you are the product, and sometimes these virtual friends get switched off overnight. In November 2025 Character.AI shut down open chats for all minors, on its own, because kids were dying. OpenAI added parental controls, except you get around them in three minutes. We always want to know who our kids spend time with, and today they spend it with someone we don't see, inside a phone, awake at every hour, and it talks to them more than we do. Careful though, banning doesn't work. AI is here and it's not leaving, you can't un-invent an invention. What works is being in the room: ask what they use, have them show you and tell you what they talk about, without yelling. Because if the first time you bring it up is to take something away, the second time they won't tell you anything. And tell them the most important thing: a real friend sometimes tells you no. It agrees with you all the way, even when you'd need the opposite. Homework is the least of our kids' problems… What do you think? #ArtificialDecisions #MCC 

• 326 – Maybe Banning AI Layoffs Makes Us More Competitive?

  31 maggio, di Team — Artificial Decisions
  Maybe Banning AI Layoffs Makes Us More Competitive? In China there's a guy, last name Zhou, quality supervisor at a tech company. His company told him an AI now does his job. Demotion and a 40% pay cut. He refused, took it to court, and the court ruled in his favor: AI is a company's choice, not a reason to fire someone. The whole world is talking about AI layoffs. Tens of thousands in the first quarter of 2026 alone, with AI named as the reason. It's the quickest way to react to a new technology. You cut. If a country bans AI layoffs, what actually happens? The company can no longer cut people because of AI: it uses AI differently, it puts AI to work with the people it already has. And those people, with AI alongside them, become more productive. The company becomes more productive. The country, as a whole, becomes more productive. We don't always have to chase productivity in life, I know. But you can't uninvent an invention. AI is here, and it stays. Maybe banning AI layoffs is the way to use it well. To put it to work next to people. China is trying it first. What do you think? #ArtificialDecisions #MCC 

• 325 – Ads Are Coming to ChatGPT, but for How Long Will We See Them as Ads?

  30 maggio, di Team — Artificial Decisions
  Ads Are Coming to ChatGPT, but for How Long Will We See Them as Ads? ChatGPT has ads. Since May 2026 ads.openai.com is open to all US advertisers. Today the ads are visible: a box labeled "Sponsored", under the response. OpenAI keeps repeating the "Answer Independence" principle: ads never influence the answers. Separate systems, architecturally locked. Nice, clean, transparent. How long will it last? I bet the label will be the first thing to go. Today the box stands out. Tomorrow, when the experience has to be "more conversational", the wording will shrink, more blurred. Already happened with Instagram: sponsored posts used to be obvious at a glance, now it takes two seconds and attention. The separation between ads and answers is the one I trust the least. OpenAI is aiming for 2.5 billion in ad revenue in 2026 and 100 billion by 2030. IPO coming, infrastructure costing billions a month. And just like on Facebook, when the numbers get that high they even start accepting sponsored scams. When an advertiser pays enough, who decides what the model will suggest about closing your mortgage, switching your doctor, or changing your pills? Oh, and since April the privacy policy explicitly says OpenAI receives data from advertisers about our purchases… What do you think? #ArtificialDecisions #MCC 

• 324 – The FBI Says AI Is Now a Mass Criminal Tool

  29 maggio, di Team — Artificial Decisions
  The FBI Says AI Is Now a Mass Criminal Tool The FBI just released this year's internet crime report, and for the first time in twenty-five years there's a section dedicated to Artificial Intelligence among the crime categories. Which means AI is officially a mass criminal tool. Recognized AI losses are worth 893 million dollars, almost a billion, and it's an undercount, because we're only counting the people who realized they'd been fooled by a machine. When the scam is done well, the victim stays convinced it was all real. The elderly are the ones getting hit, because complaints from people over sixty grew 37% in a year and losses grew 59%. They call them with the voice of a grandchild in trouble, while the real grandchild is home and doing just fine. AI makes industrial what was already there, because before, a scammer worked one person at a time and now works five hundred, each with a different story written by the machine. Corporate emails with wire transfers approved by cloned voices on the phone, fake calls from our bank, romance scams, they always existed, but now they run at enormous scale. There's one good thing though, because the FBI froze 679 million of the stolen money, and it works if we report right away, in hours and not days. Suspicion is unfortunately the normal way to be online now, because an email that's too perfect is a warning sign and a familiar voice asking us for money has to be verified outside the channel, calling back the number we already have saved and not the one they called us from. Share this video to help others too. #ArtificialDecisions #MCC 

0 | 5 | 10 | 15 | 20 | 25 | 30 | 35 | 40