Founder, HumaLitix🔹Data & AI🔹I/O Psychology & Behavioural Science🔹Transforming Organisations with People-Centric Data and AI adoption solutions
The next quantum leap in AI may not be imminent, and that is a good thing! When LLMs broke into the scene, the quantum leap in GenAI took most of us by surprise. And alongside all the buzz and hype, there were claims that AGI could be in by as early as 2026. But beyond the hyped halls of social media, as experts across the globe sifted through LLMs, conducted real experiments, published papers, started blogging and speaking, a consensus has started to emerge - LLMs do not reason! And GenAI is not a sufficient ingredient for AGI. Of course, there will be interesting domain specific applications that will solve pseudo-reasoning tasks as well, by combining with RAG, through agentic orchestrators etc. But I am definitely leaning towards the school of thought that AGI will need another quantum leap that has not been achieved yet, and it is not imminent! What is more likely, and will create significant business value meanwhile, is that compound systems - consisting of a combination of traditional rule-based software, Classical ML models, GenAI, RAG, Agentic workflows etc will gradually be adopted *widely* and matured *gradually* in useful real world applications beyond generating text, images and video. In the process, we will learn step by step how to use AI in wide real world application, where and why a human in the loop is necessary etc. What do you think? #artificialintelligence #learning
