AI versus Marshy - The Individual versus the Business Gap in AI

I’ve built a career in and around digital marketing, but I’ve learned to avoid mentioning it because it often leads to a predictable conversation. I’ve worked with Google and Facebook, but I’ve found that talking about them can derail the conversation in a way I don’t want. Let’s talk about something else - like AI. But most of the time I’ll avoid dropping that term because do you know what happens when you do? Everyone’s an expert. I’ve heard the whole freaking thing: “Oh I never click on those ads they don’t work” “You mean those things that follow me around” “Can you have a look at my nephew’s FB page” I just avoid mentioning it altogether (and Google and Facebook) because it railroads the conversation in a direction I usually don’t want it go in. Now picture AI. Nearly everyone in the Western world has heard about it and most have at least looked at it. There’s a smaller but sizeable cohort who are using it every day, and then there’s a subset that I would describe as power users. Even within this there’s a range - moving from “I have multiple LLM subscriptions” to “I self-host my own LLM and have trained to dominate my niche”. This presents a dilemma to forward-facing leaders of big businesses. They can see what’s possible. “Look - I’ve seen all of this information do this ”. But the reality is big business hasn’t taken to AI adoption like those power users I’ve mentioned because it’s much harder . My friend Nic wrote about this a couple of weeks ago where most people are finishing with a copy + paste. Check out Nic’s sometimes newsletter! We’re in this weird tech adoption curve where individuals are running at lightspeed, and companies (and the companies selling services in this space) are scrambling to figure it out. I have a LOT of thoughts on this, here are a few: Businesses don’t move fast. The adoption of mobile, social media, and the Internet were all revolutionary but took 10+ years to truly take hold. Generative AI is going two orders of magnitude plus faster than this but I see it popping offer within 5-10 (we’re 3.5 years in). Context is everything. Data needs to be robust. I am yet to work with a company that has laser-level data hygiene. It’s never been a top priority and most businesses run anywhere from none to 60-70% is close enough (Meta and Google included). Data atrophies. Records become old, less useful, and more meaningless over time. We understand this contextually as humans and make informed decisions with incomplete information all of the time - this nuance is nigh-on impossible to replicate by a machine right now. Does it mean it doesn’t work at all though? No - in the last 3-4 weeks I’ve spoken with three different Australian business owners who don’t know each other and are billing 5, 6, and 7-figure deals orchestrating real solutions for such businesses. The challenge - and why some people read these words - is filtering out the bulls**t from the value. It’s there. Just hiding under a lot of people who understand AI because they’ve played with ChatGPT. Generating video ads versus strategising video ads Remember the first time you created an image with AI? It felt mind-blowing right? Then we saw image after image after image and things wore off after a while. I’m not that impressed by the VEO demos - not because they’re not jaw dropping, but because it’s Google. ​ It’s incredible but not creative. You have access to every YouTube of all-time - I would expect the model to be stupid good and the prompts don’t seem like a super creative use case. I’m seeing a natural tension emerging between new world and old world - and this couldn’t have been painted more clearly than this exchange I witnessed between a “programmatic video creator” and my friend Andrew: Originally appeared in newsletter : AI versus Marshy 62 - the tension between old and new modes