Futurice

More good research coming out last week about sycophantic AI : https://lnkd.in/e63XYRRD . Two key findings that jumped out to me: ▪️ Participants in the study significantly preferred a sycophantic AI chatbot over alternatives designed to be neutral or challenging. They said it understood them better, was more helpful and they were more likely to act on its advice. ▪️ After experiencing the 'unconditional affirmation' of sycophantic AI, seeking advice from humans felt more effortful and less needed. For future advice, participants said they were almost as likely to go to the AI as close family and friends. This adds more evidence to what I was writing about last week, on the dangers of working with a sycophant (aka today's AI): https://lnkd.in/eMdg_fbx When we intuitively prefer sycophantic AI, many of us will subconsciously avoid pushing AI to challenge us. And we can't rely on AI labs to fix the problem, when it means creating a product their users will like less. At the same time, this research shows that the frictionless feeling of validation we get from AI feedback makes us less likely to seek real feedback from humans. So, I reckon more people should be thinking about how to embed challenge and deep thinking into their AI workflows - to pop our cosy AI bubbles.

The best part in mentoring or job search help/coaching is meeting young talent. I get to meet talented individuals and no people are alike despite sharing similar challenges. Many are straight out an example and inspiration to me. Tenacity and perseverance are bursting through seams. Check out e.g. Muskan Dangol who out of struggle with job search developed something useful for others, while also using this opportunity to learn new skills and grow. Hei employers out there, take a note of this talent! 👇🏻👇🏻👇🏻