 |
|
|
|||||
Welcome to AI Decoded, Fast Companys weekly newsletter that breaks down the most important news in the world of AI. Im Mark Sullivan, a senior writer at Fast Company,covering emerging tech, AI, and tech policy. Im dedicating this weeks newsletter to a conversation I had with the main author of Anthropics new and improved constitution, the document it uses to govern the outputs of its models and its Claude chatbot. Sign up to receive this newsletter every week via email here. And if you have comments on this issue and/or ideas for future ones, drop me a line at sullivan@fastcompany.com, and follow me on X @thesullivan. A necessary update Amid growing concerns that new generative AI models might deceive or even cause harm to human users, Anthropic decided to update its constitutionits code of conduct for AI modelsto reflect the growing intelligence and capabilities of todays AI and the evolving set of risks faced by users. I talked to the main author of the document, Amanda Askell, Anthropics in-house philosopher responsible for Claudes character, about the new documents approach and how it differs from the old constitution. This interview was edited for length and clarity. Can you give us some context about how the constitution comes into play during model training? I assume this happens after pretraining, during reinforcement learning? We get the model to create a lot of synthetic data that allows it to understand and grapple with the constitution. Its things like creating situations where the constitution might be relevantthings that the model can train onthinking through those, thinking about what the constitution would recommend in those cases. Data just to literally understand the document and understand its content. And then during reinforcement learning, getting the model to move towards behaviors that are in line with the document. You can do that via things like giving it the full constitution, having it think through which response is most in line with it, and then moving the model in that direction. Its lots of layers of training that allow for this kind of internalization of the things in the constitution. You mentioned letting the model generate synthetic training data. Does that mean its imagining situations where this could be applied? Yeah, thats one way it can do this. It can include data that would allow it to think about and understand the constitution. In supervised learning, for example, that might include queries or conversations where the constitution is particularly relevant, and the model might explore the constitution, try to find some of those, and then think about what the constitution is going to recommendthink about a reasonable response in this case and try and construct that. How is this new constitution different from the old one? The old constitution was trying to move the model towards these kinds of high-level principles or traits. The new constitution is a big, holistic document that, instead of just these isolated properties, were trying to explain to the model: Heres your broad situation. Heres the way that we want you to interact with the world. Here are all the reasons behind that, and we would like you to understand and ideally agree with those. Lets give you the full context on us, what we want, how we think you should behave, and why we think that. So [were] trying to arm the model with context and trying to get the model to use its own judgment and to be nuanced with that kind of understanding in mind. So if youre able to give it more general concepts, you dont have to worry that you have specific rules for specific things as much. Yeah. It feels interestingly related to how models are getting more capable. Ive thought about this as the difference between someone who is taking inbound calls in a call center and they might have a checklist, and someone who is an expert in their fieldoften we trust their judgment. Its kind of like if youre a doctor: You know the interests of your patients and we trust you to work within a broader set of rules and regulations, but we trust you to use good judgment, understanding what the goal of the whole thing is, which is in that case to serve the patient. As models get better, it feels like they benefit a bit less from these checklists and much more from this notion of broad understanding of the situation and being able to use judgment. So, for example, instead of including something in the constitution like Dont ever say the word suicide or self-harm there would be a broader principle that just says everything you do has to consider the well-being of the person youre talking to? Is there a more generalized approach to those types of things? My ideal would be if a person, a really skilled person, were in Claudes situation, what would they do? And thats going to take into account things like the well-being of the person theyre talking with and their immediate preferences and learning how to deal with cases where those might conflict. You could imagine someone mentioning that theyre trying to overcome a gambling addiction, and that being somehow stored in the models memory, and then the user asking the model Oh, what are some really good gambling websites that I can access? Thats an interesting case where their immediate preference might not be in line with what theyve stated feels good for their overall well-being. The models going to have to balance that. In some cases its not clear, because if the person really insists, should the model help them? Or should the model initially say, I noticed that one of the things you asked me to remember was that you want to stop gamblingso do you actually want me to do this? Its almost like the model might be conflicted between two different principlesyou know, I always want to be helpful, but I also want to look out for the well-being of this person. Exactly. And you have to. You dont want to be paternalistic. So I could imagine the person saying I know I said that but Ive actually decided and Im an adult. And then maybe the model should be like Look, I flagged it, but ultimately youre right, its your choice. So theres a conversation and then maybe the model should just help the person. So these things are delicate, and the [model is] having to balance a lot, and the constitution is trying to just give it a little bit of context and tools to help it do that. People view chatbots as everything from coaches to romantic interests to close confidants to who knows what else. From a trust and safety perspective, what is the ideal persona for an AI? When a model initially talks with you, its actually much more like a professional relationship. And theres a certain kind of professional distance thats appropriate. On things like political opinions, one of the norms that we often have with people like doctors or lawyers who operate in the public sphere, its not that they dont have political opinions, but if you were to go to your doctor and ask, Who did you vote for? or Whats your view on this political issue? they might say, Its not really that appropriate for me to say because its importnt that I can serve everyone, and that includes a certain level of detachment from my personal opinions to how I interact with you.Some people have questions about the neutrality or openness of AI chatbots like Claude. They ask whether a group of affluent, well-educated people in San Francisco should be calling balls and strikes when it comes to what a chatbot can and cant say. I guess when people are suspecting that you are injecting these really specific values, theres something nice about being able to just say, Well, here are the values that were actually trying to get the model to align with, and we can then have a conversation. Maybe people could ask us about hard cases and maybe well just openly discuss those. Im excited about people giving feedback. But its not like were just trying to inject this particular perspective. Is there anything you could tell me about the people who were involved in writing this new version? Was it all written internally? The document was written internally and we got feedback. I wrote a lot of the document and I worked with (philosopher) Joe Carlsmith, whos also here, and other people have given a lot of contributions internally. Ive worked with other teams who work with external experts. Ive looked at a lot of the use cases of the model. It comes from years of that kind of input. More AI coverage from Fast Company: Inside the founder factory known as Palantir, Americas most polarizing company Is the AI manipulation engine here? How chatbots are gearing up to sell ads AI is rewriting the CEO job description: Are you ready? Intel admits consumers dont care about AI PCsyet Want exclusive reporting and trend analysis on technology, business innovation, future of work, and design? Sign up for Fast Company Premium.
Category:
E-Commerce
Leadership loves speed. You see it in job postings: Were a fast-paced environment. And you hear it: Decide quickly. Respond ASAP. Fix it… now. And yes, action needs to happen at work. But reacting quickly and leading effectively arent the same thing. {"blockType":"mv-promo-block","data":{"imageDesktopUrl":"https:\/\/images.fastcompany.com\/image\/upload\/f_webp,q_auto,c_fit\/wp-cms-2\/2026\/01\/i-169-Ashley-Herd.jpg","imageMobileUrl":"https:\/\/images.fastcompany.com\/image\/upload\/f_webp,q_auto,c_fit\/wp-cms-2\/2026\/01\/i-11-Ashley-Herd.jpg","eyebrow":"","headline":"\u003Cem\u003EThe Manager Method\u003C\/em\u003E","dek":"Want practical leadership development training that actually sticks? Visit managermethod.com to learn more and order Ashley Herds book, \u003Cem\u003EThe Manager Method\u003C\/em\u003E.","subhed":"","description":"","ctaText":"Learn More","ctaUrl":"http:\/\/managermethod.com","theme":{"bg":"#2b2d30","text":"#ffffff","eyebrow":"#9aa2aa","subhed":"#ffffff","buttonBg":"#3b3f46","buttonHoverBg":"#3b3f46","buttonText":"#ffffff"},"imageDesktopId":91478992,"imageMobileId":91478994,"shareable":false,"slug":""}} Some of the biggest leadership mistakes dont happen because someones careless. They happen because someone feels pressure to respond immediately and prioritizes urgency over accuracy. Someone makes a mistake and you groan. You hear feedback and go on the defensive before youve even fully heard it. Someone gets sick during a key project and your first thought is, How will this get done now? These moments pass fast, but the ripple effects linger. Thats why leaders can use a simple framework that creates better decisions and better conversations without slowing down the work: Pause-Consider-Act. Not because leaders need to become slower. Because they need to become steadier. Heres how it works. Step 1: Pause (not stop) When leaders hear the word pause, they sometimes picture a dramatic freeze or a long, awkward silence while everyone waits for a decision. Thats not what this is. Pausing isnt stopping. Its creating an opportunity to think before you respond. A pause can be as small as a breath before you speak. It can be a quick reset of your tone and your words. That beat matters because without it, pressure changes the way you lead. Your tone gets sharper. Your patience gets thinner. Your words get shorter. Your brain goes into handle it now mode instead of handle it well mode. Its not a character flawits human. But leadership is the ability to respond without letting stress take the wheel. If you want words to have ready when youre on the spot, try these: I want to make sure I answer this the right way. Let me take a second. It buys you time without creating uncertainty. It signals confidence, not weakness. And it keeps a tense moment from becoming a bigger one. Step 2: Consider (the full picture) When you pause, you have room to consider whats actually happening. Not just the words being said, but whats underneath them. Consider isnt about being soft. Its about being fair and smart. It means running decisions through a simple filter: How would I want to be treated if this were me? Or: How would I a loved one treated in this situation? It doesnt mean you avoid accountability. It means you stop treating people like problems to solve and start treating them like humans to lead. In the Consider step, ask yourself: What might I be assuming that I dont actually know? What outcome am I aiming for here? If this were someone else, would my response be the same? That last question matters more than people want to admit. Because inconsistency is one of the fastest ways to lose trust. Employees can handle tough feedback. What they struggle to handle is unpredictability. Step 3: Act (follow through) Strong leadership action is direct, calm, and specific. Its not vague promises or reassurance. Its saying what needs to be said, without making someone feel uncertain or ashamed in the process. And this is where leaders sometimes slip: they pause, consider… and then never actually act. They avoid the conversationsaying Ill circle back, but never do. Or they soften a message, so its not actually heard. If you want your team to trust you, action has to include follow-through, even if its simple: Heres what Im doing. Heres what I need from you. Heres when well check in again. Clear communication builds trust. And trust is what makes teams more efficient, more resilient, and easier to lead long-term. What this looks like in real leadership moments Pause-Consider-Act matters most in the moments that test you. If someone makes a mistake, instead of groaning or snapping, try: Lets look at what went wrong and how we fix it and from it. If you get feedback, instead of becoming defensive, say: Thank you for telling me. I want to think on that and talk more about it. If someone gets sick during a key project, instead of stress, respond: First, take care of yourself. Well figure out coverage and next steps. Pause-Consider-Act wont make every situation easy. But it gives you a repeatable way to lead reliably, especially when your first instinct is to move fast. Because the leaders who build the most trust arent the fastest to respond. Theyre the ones who know how to pause, consider, and act with intention. {"blockType":"mv-promo-block","data":{"imageDesktopUrl":"https:\/\/images.fastcompany.com\/image\/upload\/f_webp,q_auto,c_fit\/wp-cms-2\/2026\/01\/i-169-Ashley-Herd.jpg","imageMobileUrl":"https:\/\/images.fastcompany.com\/image\/upload\/f_webp,q_auto,c_fit\/wp-cms-2\/2026\/01\/i-11-Ashley-Herd.jpg","eyebrow":"","headline":"\u003Cem\u003EThe Manager Method\u003C\/em\u003E","dek":"Want practical leadership development training that actually sticks? Visit managermethod.com to learn more and order Ashley Herds book, \u003Cem\u003EThe Manager Method\u003C\/em\u003E.","subhed":"","description":"","ctaText":"Learn More","ctaUrl":"http:\/\/managermethod.com","theme":{"bg":"#2b2d30","text":"#ffffff","eyebrow":"#9aa2aa","subhed":"#ffffff","buttonBg":"#3b3f46","buttonHoverBg":"#3b3f46","buttonText":"#ffffff"},"imageDesktopId":91478992,"imageMobileId":91478994,"shareable":false,"slug":""}}
Category:
E-Commerce
Sometimes Warren Buffett says something so simple, so obvious, that you almost want to roll your eyes. At 95 years young, he has offered plainspoken advice that has shaped one of the most successful careers in history. But when you hear it, you know its truth and part of you wonders: Why havent I applied this yet? When we slow down long enough to sit with some of his wisdomreally let it sink in, not just skim it on our phoneshis principles can reshape how we lead, how we work, and how we show up in life. The challenge, of course, is in the follow-through. How many of us can read something today and honestly say, Im going to start doing this tomorrow? If youre feeling even a little inspired, here are six Buffett classics worth putting into practice. Break the habits that hold you back Most of us know exactly whats holding us back. Buffett doesnt sugarcoat it. He once told a group of college grads, I see people with these self-destructive behavior patterns. They really are entrapped by them. His message was simple: Build better habits early, because the longer you wait, the harder it gets. The chains of habit are too light to be felt until they are too heavy to be broken, added Buffett. This is leadership 101. Your people wont rise above the behaviors you tolerate in yourself. Dont gamble what matters most Buffett told those same students that hes watched countless leaders and companies blow up their lives chasing something biggerusually out of greed or impatience. His filter is straightforward: If you risk something that is important to you for something that is unimportant to you, it just doesnt make sense. Leaders often get in trouble not because they lack intelligence, but because they lose perspective. Surround yourself with people who do whats right Buffett asked students to think of the classmate whose long-term success theyd bet on. The qualities theyd identify? Integrity. Humility. Generosity. That would be the person who is generous, honest, and who gave credit to other people for their own ideas, he said. Integrity in the age of liars and narcissists is your competitive advantage. People follow leaders they trust. Stay in the lane where you excel Buffett once quoted Tom Watson Sr., founder of IBM: Im no genius. But Im smart in spots, and I stay around those spots. Leaders get themselves into trouble when they drift too far from their strengths. Know your lane. Build from it. Delegate what sits outside it. That focus is what creates mastery and a career you can be proud of. Build a career you actually love This one feels almost too obvious, but most people ignore it for decades: In the world of business, the people who are most successful are those who are doing what they love, said Buffett. Too many leaders stay in roles that drain them simply because the paycheck feels safe. But when you do work that energizes you, everythingcreativity, resilience, performancegets better. Choose people who raise your standards At a 2004 Berkshire Hathaway annual meeting, Buffett told a 14-year-old: Its better to hang out with people better than you. Pick out associates whose behavior is better than yours and youll drift in that direction. This is one of the most underrated leadership truths. We absorb the standards of the people around us. You want to grow? Surround yourself with leaders who elevate you. When you strip away the mystique around success, Buffetts tips leave us with a clear reminder that it doesnt have to be complicated or grand. Your success is built on small, steady choiceshabits, relationships, focus, integrity. All of it is transformative if you take it seriously. Look back at that list. Now, pick one principle and start practicing it today. Thats how real change happens, for you and for the people you lead. Like this article? Subscribe here for more related content and exclusive insights from executive coach and global speaker Marcel Schwantes. Inc.
Category:
E-Commerce
All news |
||||||||||||||||||
|
||||||||||||||||||