C_NCENTRATE 914: Polycrisis, generative AI, Meta scraping, suicide markers, mixed reality +
WE NEED TO TALK ABOUT GENERATIVE AI
Unless you’ve been under a rock, generative AI is going to make you redundant, completely charge marketing, kill us all, and a lot more besides. OpenAI has developed the most famous model so far, ChatGPT, a state-of-the-art language model. GPT stands for ‘Generative Pre-trained Transformer’, and using a transformer neural network (not the cartoon robots variety), the program can generate human-like text.
Sound sinister? Yes and no. The model has been trained on a diverse dataset of web pages, books, and other text sources, which allows the results to offer a wide range of knowledge and the ability to generate text in various styles and formats. The tool is great for summary, translations and ‘conversation’, but the outputs are often imperfect and can take a lot of inputting and trial and error to get correct. Time and error will, of course, improve both of these things. The issue? Where to start? Universities? We’ll need a detector to stop the kids from plagiarising. Lazy marketers? They’re lining themselves up, saying this is a good reason to fire your staff (don’t do that). What about bad actors looking to flood networks with propaganda and other rubbish? It just got faster. There’s a long list of what this could be used for.
SO WHAT?
__ DO __ Know that it’s good when well used. The program is highly effective and has scored well against multiple benchmarks. Undoubtedly this will help many people do jobs faster, increase knowledge and other areas. // __ DON’T __ Think there aren’t big limitations. ChatGPT, like other language models, can struggle with understanding context and can generate text that is biased or offensive. This is primarily due to the data it has been trained on, which can contain biases and errors. The model's large size and computational requirements can make it difficult to deploy in some situations. Ask yourself this. If your life depended on it, could you tell what pieces of this were written using ChatGPT?
If you want the real deal on how to get the most out of generative AI now and in the future, book a session with us here.
HOW TO HANDLE ANYTHING THAT COMES YOUR WAY: You can order the new version of ‘Disruptive Technologies’ containing the TBD framework.
If you’re in e-commerce, retail, FMCG, healthcare, technology or content, you need to subscribe to ‘What Did Amazon Do This Week?’. Everything Amazon does, every week. Featured in the Wall Street Journal and called ‘Obsessively detailed’. Subscribe now (£15pm/£100pa) and tell a friend; they’ll thank you.
/// USE: Sound better at work. [SoundBetter]
/// READ: ‘Disruptive Technologies V2’ by This Moi. [Amazon]
/// LISTEN: ‘Start with This’ [Start With This]
Could genetic prevention be possible for suicide? /6 mins
Paying people to have vaccines doesn’t dent people’s trust in them. /6 mins
Why Meta is (and isn’t) driving the Metaverse narrative. /6 mins
Are suburbs about to bid for residents to secure their future? /4 mins
CNET has been using AI-written articles for months. /5 mins
Microsoft has developed a bot that can copy your voice after 3 SECONDS. /4 min
The war on cash might not be going as well as some want. /3 mins
Meta is suing “scraping-for-hire” service that sells your user data to law enforcement. /5 mins
How mixed reality works. /7 mins
Cancer vaccines are showing promise. /5 mins
We’re in a ‘polycrisis’ era, according to the WEF. /6 mins