Gary Jones

When the internet first emerged, many were sceptical about its potential for monetisation. It was a vast, uncharted territory, and the idea of making money online seemed far-fetched. However, as the internet evolved and its capabilities expanded, innovative minds found ways to capitalise on this new frontier, giving rise to e-commerce, online advertising, and many of the other revenue streams, that we now take for granted. Today, we find ourselves in a similar situation with the rapid advancement of OpenAI and Large Language Models (LLMs). These powerful technologies have captured the imagination of the world, but the question of how to effectively monetise them remains a puzzle. Just as the internet disrupted traditional business models, LLMs are poised to revolutionise entire industries, from content creation to customer service, personalised recommendations and beyond. I feel so excited to be working closely with a global team of extraordinary entrepreneurs and visionaries, who are now exploring the vast potential of LLMs and new monetisation strategies. The clock is ticking and those who can harness their power effectively will undoubtedly reap the rewards. If you are looking for a way to make money with AI, we need to talk! My team of global VISIONEERS is growing fast. We are embracing innovation and unlocking incredible strategies that will change the way you work and think forever. If you are open to embracing these technologies and hungry to learn how to monetise them, message me and we will set up a time for a virtual meeting and I will show you some of the incredible things we are creating that could change your future forever. #chatgpt4o #AI #LLM #thefutureisnow #futurist

2

1 Comment

Do you want to know if your copyright data was used to train AI models? A team at Imperial College London developed “copyright traps” to help you with that. 🛠️ HOW IT WORKS? “Copyright traps” are hidden text that can be inserted into content to later detect if it was used in AI training datasets. The idea is to generate gibberish sentences and inject them into text. If an AI model shows low “surprise” when encountering these sentences, it suggests the text was part of its training data. The team successfully tested their traps on CroissantLLM, a 1.3 billion-parameter model - much smaller than giants like GPT-4 (est. 1.76 trillion parameters). 🔍 LIMITATIONS The “copyright traps” are not perfect. They require lots of text alteration, making it detectable and potentially removed by AI trainers. 🔄 CONCLUSION While not a perfect solution, the “copyright traps” are a new step further in the ongoing debate over AI training data and creator’s rights. What do you think? Could this reshape content protection in AI? #ArtificialIntelligence #IntellectualProperty You can read more here:

2