Maria Sukhareva | On AI slop and AI experts, and the sleep of colourless green ideas | Episode #94

In this episode of The Only Constant, Lasse Rindom speaks with Maria Sukhareva - computational linguist, Lead AI Strategist, and founder of AI Realist - who brings 15+ years in natural language processing to a refreshingly hype-free look at what large language models actually are, and aren't. With roots in computational linguistics and machine translation, Maria has been doing language modelling "since before it was cool," and she puts that depth to work as the conversation moves from the latest AI market drama into the deeper questions of language, intelligence, and truth. Main topics they discuss include: Why understanding language matters for understanding LLMs, and how the field drifted from linguistics toward pure statistics The absence of real concepts, reasoning, and truth in models that simply predict the next token AI slop, eloquence-as-intelligence, and the flood of machine-generated academic papers Why "what to do" now matters more than "how to do it" - and what this means for experts, managers, and the future of work If you've ever wondered what's actually happening behind the curtain of a language model - this conversation is for you. --- Do you want to know more about Maria Sukhareva? Maria Sukhareva is the founder of AI Realist (airealist.org), an independent publication offering vendor-neutral, anti-hype analysis of enterprise AI for a practitioner audience. Through her weekly AI Realist Radar briefings, long-form investigative writing, regular workshops, and keynote engagements, she covers the developments enterprises actually need to navigate, from model releases and pricing shifts to security incidents and governance frameworks. AI Realist has grown to over 30,000 followers on LinkedIn and hundreds of paying subscribers, with her work cited by top media outlets including Forbes and Die Welt. Alongside AI Realist, Maria serves as Principal Expert in AI and Lead AI Strategist at a DAX40 company, with over 15 years in NLP and generative AI. Her path runs from certified translator to computational linguist at Universität des Saarlandes, through leading the machine translation initiative at BMW Group, to her current role shaping enterprise AI strategy. Maria holds a background in computational linguistics, has authored more than 20 publications, and speaks four languages.