Toju Duke

While emotional chatbots and AI companions are seemingly empathetic and supportive, they’ve led people down some very dark and ugly sides. So if you’re using LLMs for emotional support, proceed with caution, limit the amount of time spent and your dependency on them.

If you outsource 90% of your work to Gen AI, it does all your thinking for you. So rather, outsource 10-30% of your work to LLMs and use 70% of your time to ideate, create, plan and do the tasks you absolutely love.

I remember giving a talk to marketing professionals on AI way back in 2022, just before the launch of ChatGPT. The main takeaway was human intelligence and human expertise are still relevant; this is even more critical and true in today’s world of Gen AI and AI agents. I’ve delivered keynotes at different marketing events from various industries over the years, and one thing has always stood out - marketers are so good at what they do, so quick to learn, and are one of the warmest and most engaging audiences. Considering I worked in marketing for a few years - a part of my career I sometimes forget - it’s one field we can’t do without. So on this note, I’m absolutely delighted to be delivering a keynote at eCommerce Expo and Technology for Marketing on 23 September at Excel London. I’ll speak on AI Governance for eCommerce: The Risks, Red Flags and Rules That Matter Key takeaways include: ✔ The biggest AI mistakes eCommerce and retail brands should avoid ✔ Where governance, safety and human oversight matter most in customer-facing AI ✔ How to drive personalisation and not automation  ✔ Why transparency, explainability and accountability are becoming commercial advantages ✔ A practical framework for using AI in ways that are effective, trusted and commercially sustainable And that’s not all, Diverse AI is exhibiting so we’ll be there to make new friends, form partnerships, and share our amazing work. Register for free: https://lnkd.in/dF-xzxrd

Will China move up the ranks and become the latest AI superpower given Z.ai’s recent release of GLM-5.2, a worthy competitor to Mythos 5? With the constant lockdowns from the US government on new frontier AI models and the cheap open-source ones from China, will businesses move their tech stacks to Chinese models? There's still a lot to consider as the US’s dominance on AI models is top of mind. Read more in the latest edition of “AI Reflections”. 👇🏼

We’ve been using AI for a long time, long before ChatGPT. For example, using facial recognition to unlock your phone or Netflix personalisation to find your favourite series uses AI. We mustn’t conflate the issues with Gen AI and traditional AI. AI is still very useful, and will be fixed.

I never considered stammering a disability or a neurological disorder. To be fair, stammering isn’t something I’ve thought about recently. But did you know that 166 million people, roughly 2% of the world’s population, stammer? And in the UK, it’s 3% of the population, equating to 1.5 million adults. That’s A LOT of people. A lot of people who face or have suffered from: ✔️ Social anxiety ✔️ Workplace Bullying  ✔️ Depression ✔️ Shame and frustration ✔️ Underemployment And many other challenges. But this is where it gets better. There are several AI solutions that help people who stammer, such as AI tutoring that reduces the severity of stuttering and social anxiety particularly amongst English learners, and helps improve emotional safety. We also have AI assisted speech therapy, improving voice-to-text using Automatic Speech Recognition (ASR) and offering real-time voice conversion to reduce severely stuttered speech, making communication much easier. That’s why I’m so pleased to be speaking at STAMMAFest on Sat, 22 Aug at 9.30am, where I’ll discuss the different ways AI can assist people who stammer, recent work carried out in this area, ways people can use AI to assist with the disability, and general tips on how to live and work with AI. I’ll be speaking with people who are doing outstanding work in this field - Richard Cave PhD from UCL, Shaomei Wu founder of AImpower.org and Jane Powell, CEO of STAMMA, event host and a UK trailblazer helping people who stammer. Dates: 20 - 23 August Venue: University of Northampton If you belong to the stammering community or you’d love to learn more about the disability, you’re more than welcome to join us. Ticket sales end in 4 days! Book your tickets here: https://stammafest.org/

From policy analysts, to HR professionals, researchers, students, educators, business executives and many others, AI summarisation tools are one of the best use-cases of generative AI and have proven efficiency and productivity, when critical thinking skills have been applied.

It’s time to up your AI game! If you’re a working professional without adequate AI governance or Responsible AI knowledge, then you’re not an AI expert. Yes, you read that right. How can you work with AI if you’re totally in the dark on its risks to you, your business and organisation? What would you do if you’re faced with a regulatory fine or class-action lawsuit due to AI? Aka AI liability and accountability. What’s the best way to procure, set-up and work with AI using the most suitable AI strategies? That’s why I’m introducing the very first of many more masterclasses - A Non-Technical Guide to Responsible AI. Join me on Tuesday, 14 July at 3pm GMT for a 90 minute fun, interactive and highly educational Masterclass on Responsible AI. Featuring hands-on exercises and a live Q&A, we’ll discuss different key concepts like AI ethics and safety and review eight distinct dimensions of AI risk. We’ll also review international regulation and operational frameworks like NIST AI RMF and ISO/IEC 42001. You’ll learn how to navigate evolving global compliance landscapes, including regulations across the EU, US, China, and Singapore, and more. Remember I said this is the first of many? Next up is a Masterclass on AI workflows for administrative roles and a full online course - more on this to come soon! Don't forget, you’re the captain of your life and career. Steer it in the right direction. More details and link to register: https://luma.com/py2xbs7d Early bird tickets end in 7 days! Hope to see you there!

The basic principles of AI development will never be obsolete. The need for AI policies, data governance, and humans-in-the-loop (HITL)—especially in the new age of autonomous AI agents—is not just a responsible AI requirement; it's an essential part of the ML life-cycle that every organisation should adopt.

Your business could go bankrupt if you have AI-related issues. Particularly user-facing issues relating to harms, decline in trust and customer / investor confidence, lawsuits and regulatory fines. No one wants that to happen. That’s why having responsible AI programmes is not an option - it's a requirement.