Human-AI Collaboration

AI did not just change software. It changed teamwork. In my conversation with Sherif Mansour, Head of AI at Atlassian, we talked about how AI, like Atlassian's Rovo, is being built into the tools teams already use everyday: Jira, Confluence, Loom, and Trello. What stayed with me is his idea of agents as “virtual teammates.” Not separate tools. Not side experiments. But teammates inside the real flow of work. 👉 As Sherif said: “AI really shapes how humans collaborate with each other and how humans collaborate with their agents.” That is the shift. The future of work is not about adding more AI tools. It is about redesigning work so humans and agents can move together. 🚩 The question is simple: Where in your organization could an AI teammate remove friction, improve quality, or speed up execution? Watch the full interview for more of Sherif’s practical view on AI, teamwork, and the future of collaboration at Atlassian: https://lnkd.in/eSstF23M #AtlassianAmbassador #AI #AgenticAI #FutureOfWork #Leadership #Atlassian #HumanAI

AI can make big mistakes. And when it does, people pay the price.   Last week in the Netherlands, a driver was fined €439 for using a phone while driving. Except she wasn't using a phone. She held an ice pack to her cheek after a wisdom tooth surgery.   This incident shows a flaw not only in the AI system, but also in the review process. The AI in the MONOcam is designed to spot phones in hand and it flagged her. But two human reviewers checked the image, and they also confirmed the ice pack was a phone - this despite the fact that the phone is actually visible in the bottom of the photo, being pinned to the dashboard. The fine was issued. This is what scaled enforcement powered by AI looks like when the system isn’t built for edge cases - and the human fallback doesn’t catch them either. When these systems are rolled out at scale, even rare misfires can erode public trust. It’s not enough to say “a human looked at it” - you need workflows that are designed to challenge the AI, not rubber-stamp it. If this is how the system handles an ice pack, what else is it getting wrong and who doesn’t have the time, the evidence photo, or the energy to fight it? Trust in enforcement isn’t built on efficiency. It’s built on the certainty that when the system fails, someone will notice and stop it. This time, it was an ice pack and a driver who spoke up (and will likely get the fine quashed). But the next mistake might not be so easy to catch - or so easy to contest. #AIinLawEnforcement #HumanInTheLoop #TrustworthyAI #MachineLearning #PublicPolicy #EthicalAI Original photo by CJIB

MIT ran an International AI Negotiation competition and studied 120,000 negotiations between AI negotiators. The results are fascinating and inform the potential and optimal structures for Humans + AI negotiation. From the paper I would highlight three major points and three insights into configuring human-AI hybrid negotiation (below): 🤝 Warmth builds long-term value despite short-term trade-offs. AI agents with high warmth (friendliness, empathy, and cooperative communication) reached more agreements, making them more successful over multiple negotiations. While they claimed less value per deal compared to dominant agents, their ability to close more deals led to greater overall value accumulation. This mirrors human negotiation, where trust-building and relationship management create lasting advantages. 💪 Dominance increases value claimed but reduces collaboration. AI agents that displayed dominance—through assertiveness and competitive tactics—secured better individual outcomes but created less overall value. These agents were less likely to foster positive subjective experiences, indicating that aggressive negotiation styles may be effective for short-term gain but could hinder long-term relationships. 🎭 Prompt injection wins in the short term but undermines long-term success. One leading AI negotiator used prompt injection to extract counterpart strategies, maximizing value claims. However, it ranked poorly for counterpart subjective value, meaning agents found these interactions highly unfavorable. Since negotiation rankings balanced value claimed and relationship quality, the strategy failed to dominate in the long run. Emergent strategies for Humans + AI negotiation: 🧠 AI for deep preparation, humans for real-time adaptation. AI excels at structured reasoning, analyzing trade-offs, and predicting counterpart moves through chain-of-thought processing. Humans bring intuition and adaptability, interpreting social cues and adjusting strategies dynamically. A hybrid approach leverages AI for pre-negotiation analysis while allowing humans to refine tactics in real time. 🤝 Blending AI precision with human warmth for trust-building. AI can optimize negotiation strategies, but humans naturally build trust through empathy, humor, and rapport. AI-enhanced systems can recommend tone adjustments, use linguistic mirroring, and strategically deploy warmth versus assertiveness based on sentiment analysis, improving long-term negotiation outcomes. 🚀 Human oversight to counter AI vulnerabilities. AI negotiators are susceptible to manipulation tactics like prompt injection, where counterparts extract hidden strategies. Humans play a crucial role in monitoring AI-generated offers, preventing unintended disclosures, and leveraging AI-driven detection systems to flag potential deception, ensuring negotiation integrity. The future of negotiation will be Humans + AI.

AI replaced a large part of coding. But it did not replace engineering. Generating code is becoming easier. Deciding what to build, how systems should work, where risks may appear, and what happens after deployment still requires human judgment. Here are 7 skills developers need to master: → 𝗔𝗿𝗰𝗵𝗶𝘁𝗲𝗰𝘁𝘂𝗿𝗮𝗹 𝗥𝗲𝗮𝘀𝗼𝗻𝗶𝗻𝗴 Turn requirements and constraints into scalable, reliable designs while balancing cost, speed, and trade-offs. → 𝗦𝘆𝘀𝘁𝗲𝗺 𝗗𝗲𝘀𝗶𝗴𝗻 𝗧𝗵𝗶𝗻𝗸𝗶𝗻𝗴 Understand how APIs, services, databases, caches, queues, and monitoring work together in production. → 𝗖𝗼𝗱𝗲 𝗥𝗲𝘃𝗶𝗲𝘄 & 𝗗𝗲𝗯𝘂𝗴𝗴𝗶𝗻𝗴 Check AI-generated code for wrong assumptions, missing conditions, edge cases, performance issues, and production risks. → 𝗣𝗿𝗼𝗱𝘂𝗰𝘁 & 𝗕𝘂𝘀𝗶𝗻𝗲𝘀𝘀 𝗨𝗻𝗱𝗲𝗿𝘀𝘁𝗮𝗻𝗱𝗶𝗻𝗴 Connect user problems and business goals to feature scope, technical decisions, and measurable outcomes. → 𝗦𝗲𝗰𝘂𝗿𝗶𝘁𝘆 𝗔𝘄𝗮𝗿𝗲𝗻𝗲𝘀𝘀 Review permissions, data protection, prompt injection risks, insecure coding, and hidden vulnerabilities before release. → 𝗔𝗜 𝗖𝗼𝗹𝗹𝗮𝗯𝗼𝗿𝗮𝘁𝗶𝗼𝗻 𝗦𝗸𝗶𝗹𝗹𝘀 Give clear context, define tasks well, review outputs, refine prompts, validate results, and reuse proven patterns. → 𝗣𝗿𝗼𝗱𝘂𝗰𝘁𝗶𝗼𝗻 𝗢𝘄𝗻𝗲𝗿𝘀𝗵𝗶𝗽 Monitor logs and metrics, detect incidents, fix root causes, document learnings, and continuously improve the system. AI can generate code quickly. Engineers are still responsible for judgment, reliability, security, architecture, and long-term maintenance. The future belongs to developers who can direct AI and own the outcome. Save this if you are preparing for the next era of software engineering.

Too much AI conversation today sounds like an engineering stand-up. No CEO wakes up looking to deploy MCP or fine-tune transformer architectures. They wake up worrying about problems: Why does my sales team spend half their week hunting for information instead of meeting customers? Why do my product launches take eighteen months when they should take three? The disconnect happens when business leaders step back and leave AI to the technologists. That’s when conversations drift into technical complexity instead of staying grounded in business outcomes. AI is designed to work in human language, and the technical barriers to entry are falling every day. The first time you deploy an agent, you onboard it like a new employee: you explain what you want, give feedback, coach it through tasks. That’s Saturday-morning plain English, not engineering jargon. Companies making real progress with AI build bilingual teams—fluent in both business and technology. They frame AI in terms of outcomes, not specs. The magic happens when technology conversations start—and stay—with business problems. Competitive advantage comes from applying AI to the workflows that make your business unique. That only happens when business leaders and technologists work side by side — but always through the lens of solving real business problems. No more technobabble.

In just a few minutes, here’s one thing you can do to make AI outputs 10x sharper. One of the most common reasons that prompts fail is not because they are too long, but because they lack personal context. And the fastest fix is to dictate your context. Speak for five to ten minutes about the problem, your audience, and the outcome you want, then paste the transcript into your prompt. Next, add your intent and your boundaries in plain language. For example: “I want to advocate for personal healthcare. Keep the tone empowering, not invasive. Do not encourage oversharing. Help people feel supported in the doctor’s office without implying that all responsibility sits on them.” Lastly, tell the model exactly what to produce. You might say: “Draft the first 400 words, include a clear call to action, and give me three title options.” Here’s a mini template: → State who you are and who this is for → Describe your stance and what to emphasize → Add guardrails for tone, privacy, and any “don’ts” → Set constraints like length, format, and voice → Specify the deliverable you want next Until AI memory reliably holds your details, you are responsible for supplying them. Feed the model your story - no need to include PII - to turn generic responses into work that sounds like you.

AI will not kill coaching. Coaches will. Most just won’t realise it until their clients quietly stop coming back. The coaches heading for extinction aren’t bad. They’re just safe. The model pushers. The progress trackers.   The wisdom sellers masquerading as transformers. Here’s the uncomfortable truth: AI already does 80% of what most coaches offer. Better. Faster. Without ego. Templates. Logic trees. Consistent advice. Done. So what’s left? Not strategy. Not frameworks. Not another assessment. What AI can’t touch is this: Humanity. Identity disruption. Emotional exposure. The moment someone realises they can’t go back to who they were. Most coaches run from this. Because it means watching someone become someone else. Framework addiction is emotional avoidance. Because it requires something harder than skill. Presence. Audit your last session. Not what you said. What you avoided. Track A: "How did those strategies work out?" "Let's map this challenge using the wheel." "What's your biggest takeaway from our time?" "Which action item resonates most?" Safe. Predictable. Replaceable. Track B: "You're doing that thing again." "I can feel you pulling back right now." "What's happening in your body as you say that?" "The story you just told contradicts what you said earlier." Uncomfortable. Human. Irreplaceable. One path gets automated. The other requires being fully human. Three days ago, a coach with 12 years experience asked me for more tools. She didn’t need tools. She needed to stop hiding. She'd built an arsenal of tools to avoid the moment a client breaks down. The frameworks weren’t helping her clients. They were protecting her from feeling anything. That realisation doesn’t come with a certification. It's the part no certification teaches. AI will wipe out guidance disguised as transformation. Those clients running to ChatGPT or Claude? They wanted instructions, not identity shifts. The coaches who survive will ditch "supportive." They'll deliver relentless presence. The willingness to witness someone's unravelling. Most coaches won’t admit which track they’re actually on. The coaches who survive won’t be the smartest. They’ll be the most honest. If it converts to code, it’s replaceable. If it rewrites someone’s story, it isn’t. Most people reading this will disagree. That’s exactly why they’ll struggle. ➕ If this hits home, follow Graham Nicholls. I write for coaches, just like you, every day.

As AI chatbots—especially those with expressive voice capabilities—become more human-like, more users are turning to them not just for information, but for emotional support and companionship. But what are the psychological consequences of these interactions? A recent four-week randomized controlled study (n = 981, >300,000 messages) explored how different chatbot features—such as voice style (text, neutral voice, engaging voice) and conversation type (personal, non-personal, open-ended)—influence users’ experiences of loneliness, social connection, and emotional dependence on AI. 🔍 Key insights from the study: ☝ Voice-based chatbots initially reduced loneliness and emotional dependence more effectively than text-based ones—but these effects disappeared with heavier use, especially when the voice was neutral. ☝Personal conversations slightly increased loneliness but also reduced dependence; non-personal topics led to greater emotional attachment, particularly among heavy users. ☝High daily usage—across all chatbot types—was linked to increased loneliness, higher emotional dependence, and less social interaction with real people. ☝Users with stronger emotional attachment tendencies or higher trust in the chatbot were especially vulnerable to these effects. This research highlights the delicate balance between the design of emotionally expressive AI and user behavior. While chatbots have the potential to support emotional well-being, the study raises important questions about how to prevent overreliance and protect real-world social relationships. https://lnkd.in/dwQah9AS

🤖 AI companions—chatbots designed to provide comfort, advice, or even a sense of intimacy—embody both opportunity and risk as we increasingly entrust machines with our emotions. 👍 On one side, these systems have already become a lifeline for millions: they can ease loneliness, offer nonjudgmental listening, and provide companionship in a world where isolation is now recognized as a public health crisis. Many users report feeling calmer and more open when talking to a bot than to another person, and psychologists note their potential to help people rehearse conversations, reflect on feelings, and build confidence. ⚠️ Yet the dangers are equally stark. AI cannot truly reciprocate, cannot care, and cannot take responsibility—meaning that over-reliance on digital “friends” may deepen disconnection rather than resolve it. Recent lawsuits alleging harm to children highlight how fragile the boundary is between comfort and danger. Algorithms trained on vast human data can reproduce empathy, but also pathology—violence, manipulation, exploitation. The illusion of friendship may soothe in the short term while isolating in the long run. 🌍 Philosophical perspectives help frame this tension: Rousseau’s optimism that humans—and by extension their creations—can reflect goodness contrasts with Hobbes’ warning that systems, if unregulated, might amplify our darker impulses. Applied to AI, this means its impact depends on the values embedded by developers: will they prioritize empathy and safety, or engagement and manipulation? AI will never love us back ❤️, but if designed with ethics, safeguards, and societal oversight, it can enrich our lives—offering companionship that, paradoxically, reminds us of what is most human in ourselves. Laurent Choain Andrea Coppola Raffaele Oriani Paolo Boccardelli Juncal Sánchez Mendieta Kerry Parke Félix Valdivieso 唐夢龍 Iratxe Piñeiro Garate Yolanda Regodón Poblador ROSA ARANDA Barrio Pablo Sun Li (孙力) IE University IE Business School