AI bubble?

November 01, 2025

by an member in Amherst, MA, USA

Wondering what everyone's take on AI is. I've been exploring its use in education. There's a lot of potential but I can't tell if it's genuinely transformative or just part of the AI bubble

0 Likes

8 Replies

139 Views

Replies

Reply by a professional
from London Business School in London, UK

I think it is potentially transformative but we have a long way to go before it reaches its potential. Because AI hallucinates (which means it makes stuff up) it can't be trusted in any area where accuracy is important. For me, accuracy is important in every area of my life and business so I therefore don't feel I can trust AI. That said, I can trust myself and my own specialist expertise, so AI can be an invaluable partner to me in the work I do. I never rely on AI to do something I don't understand or can't evaluate. But it can dramatically increase my output on things I do understand and can evaluate - e.g. my entire social media strategy is around AI - I read, analyse, research and collect my thoughts before using AI to help me organise them into content and refine that content for maximum impact. Because I know what good content looks like (I have 20 years in marketing) and because I know my subject matter I am able to do this to a high standard. Unfortunately too many people are outsourcing their thinking to AI and finding it's not working.

Reply by a professional
from University of Pennsylvania in Austin, TX, USA

I find the bubble terminology to generally not be useful, it has way too broad a set of potential interpretations. As for my opinion, you should expect AI to stick around and its value and usefulness will grow over time. Most resources are going to building better foundational models, not optimizing it for specific use cases. So even if raw capabilities cap out today, the economic impact can continue for substantially longer as resources are redirected. Even in one year of using AI directly for many work tasks, it’s been obvious that elaborate workarounds I needed in 2024 were rendered irrelevant by 2025 simply due to greater model capabilities.

+6 more replies.

Join the discussion