Changes

===LLMs===

===LLMs===

Building on earlier technologies such as neural networks, rule-based expert systems, big data, [[pattern recognition]] and [[machine learning]] algorithms, GenAI (generative AI) uses LLMs (large learning models) trained on massive data sets to create text and imagery.  

Building on earlier technologies such as neural networks, rule-based expert systems, big data, pattern recognition and machine learning algorithms, GenAI (generative AI) uses LLMs (large learning models) trained on massive data sets to create text and imagery.  

LLMs require enormous quantities of data. Existing firms in online search, sales platforms and [[social media]] platforms can exploit their own data troves. This is frequently supplemented by aggressive and unauthorised scraping of online data, sometimes confidential, leading to [[litigation]] around access, compensation and privacy. In practice, most AI models must rely on incomplete data which is difficult to clean to ensure accuracy.

LLMs require enormous quantities of data. Existing firms in online search, sales platforms and [[social media]] platforms can exploit their own data troves. This is frequently supplemented by aggressive and unauthorised scraping of online data, sometimes confidential, leading to [[litigation]] around access, compensation and privacy. In practice, most AI models must rely on incomplete data which is difficult to clean to ensure accuracy.

Despite massive scaling up of computing power, GenAI consistently fails in relatively simple factual tasks due to errors, [[bias]]es and [[misinformation]] in datasets used.  AI models are adept at interpolating answers between things within the data set but poor at extrapolation. Like any rote-learner, they struggle with novel problems. Their ability to act autonomously interacting within dynamic environments remains questionable. Cognitive scientists argue that simply scaling up LLMs based on sophisticated pattern-matching built to autocomplete rather than proper and robust world models will disappoint. Claimed progress is difficult to measure as benchmarks are vague and inconclusive.

Despite massive scaling up of computing power, GenAI consistently fails in relatively simple factual tasks due to errors, [[bias]]es and [[misinformation]] in datasets used.  AI models are adept at interpolating answers between things within the data set but poor at extrapolation. Like any rote-learner, they struggle with novel problems. Their ability to act autonomously interacting within dynamic environments remains questionable. Cognitive scientists argue that simply scaling up LLMs based on sophisticated pattern-matching built to autocomplete rather than proper and robust world models will disappoint. Claimed progress is difficult to measure as benchmarks are vague and inconclusive.

Cheerleaders miss that LLMs do not reason but are probabilistic prediction engines. A system which trawls existing data, even assuming that is correct, cannot create anything new. Once existing data sources are devoured, scaling produces diminishing returns. Rather than fully generalisable intelligence, generative models are regurgitation engines struggling with [[truth]], [[hallucination]]s and [[reasoning]].

Cheerleaders miss that LLMs do not reason but are probabilistic prediction engines. A system which trawls existing data, even assuming that is correct, cannot create anything new. Once existing data sources are devoured, scaling produces diminishing returns. Rather than fully generalisable intelligence, generative models are regurgitation engines struggling with [[truth]], [[hallucination]]s and [[reasoning]].<ref>https://www.zerohedge.com/ai/satyajit-das-ai-artificial-intelligence-or-absolute-insanity</ref>

===Open web in decline===

===Open web in decline===