Text generation is the most crowded generative AI field, but there’s a big prize for startups to win here, as plenty of people hate writing.

ChatGPT is a great brainstorm buddy. Sure, it lacks nuance, critical thinking and ethical decision-making skills — but it can easily scribe formulaic content like advertising copy, cold outbound emails, corporate training, onboarding and bits of code.

AI is on the way to mastering all manner of work admin: meeting recaps, languages translated in real-time, to-do lists based on office conversations, transcripts and call summaries.

Copyright lawyers are going to be busy chasing money from AI-generated artwork that looks like it was based on human artists’ creations.

In another way, AI will be a powerful aid for the legal industry, which relies heavily on standardised documents. These docs are a) expensive and b) a pain to put together, so it’s easy to imagine a future where they’re handled by legal language models.

Auditing firm PwC estimates that the average Fortune 2000 company has between 20k and 40k active contracts. Robot lawyers may be key to managing the workload.

Will lawmakers clip generative AI’s wings? The EU is looking to more strictly regulate the use of AI, designating some of it as “high-risk”, binding developers to requirements of transparency, safety and oversight. In February, Italy’s Data Protection Agency stepped in to prevent AI chatbot Replika from using the personal data of its users.

EU rules have already shaped AI in Europe through the GDPR privacy law (to the detriment of the field, many founders argue). Big Tech firms recognise the risk of legislation. A recent investigation by transparency group Corporate Europe Observatory said industry players, including Microsoft and Google, had lobbied EU lawmakers to exclude AIs like ChatGPT from its crosshairs.

Generative AI has replaced Web3 and crypto as the next big things — perhaps mercifully. Now many founders are rushing to emphasise the importance of AI to their business models.

We’re likely going to be subjected to lots of AI systems that are not really AI systems — so it’s best to be wary of the uncritical pluggers popping up to promote the field.