Goldman Sachs issues a report about generative AI; currently too expensive to justify ROI, no killer app and looming power & chip shortages.

That said they see money to be made if a killer app is found and also simply because bubbles take a long time to burst.

You can read the entire 31 page report below https://www.goldmansachs.com/intelligence/pages/gs-research/gen-ai-too-much-spend-too-little-benefit/report.pdf

@carnage4life the fascinating part of this take is that “the killer app” is yet to emerge.

big investment is in upstream services, LLM training & data acquisition. But none of it is successful without someone downstream having a revelation about how they can be used

more interesting, it’s not clear that we need bigger LLMs to surface that killer app. like, OpenAI & anthropic could die today and maybe AI could still flourish. that’s possible

@kellogh @carnage4life my hunch is once Apple Intelligence stuff ships and becomes an everyday tool for a large number of people, we'll start to see a bit less daydreaming about what "AI" could be (and hopefully more focus on building useful products, but who knows, maybe another hype cycle will come along)