Jay Gould is not a very well-known name. But he is one of the forces that shaped American Capitalism. He was a railroad magnate who almost brought down President Ulysses. S. Grant when he tried to manipulate gold prices. The original Black Friday was not a fun trip to a department store to bag goodies but a day that wrecked the markets due to plummeting gold prices. Jay Gould's unbridled attempts to manipulate the markets, do insider trading, and win with dubious manipulative moves led to more controls in the financial markets.

History never repeats itself. Human nature does.

ChatGPT and Open AI are new. But human nature is not.

We are in the 'Gilded Age' all over again. Instead of Railroads, it's AI. We have Sam Altman and Satya Nadella instead of Jay Gould and James Fisk. Allow me this writer's liberty, because neither Sam Altman nor Satya Nadella is as vicious as Jay Gould was but I am never the one to miss a chance to spot possible dots that could connect.

Railroads industrialized America. AI will do the same for many industries.

For eCommerce too, there are plenty of opportunities that come from the roads that AI will lay.

The future of AI is already here in eCommerce. It is just unevenly distributed.

Virtual Makeovers & trials, Labelling, and tagging of designs, styles, and SKUs are already well-established at least for the last five years. Once a human-centered activity, image recognition models are increasing both the throughput and accuracy of these activities at a rapid pace.

ChatGPT, which isn't the last of interfaces for interacting with large language models, is the toast of the town now. Trained on a vast swath of internal tagged and labeled product content as well as a wide set of comparable eCommerce websites, AI-powered chat interfaces can take over the job of a stylist and a customer support person – the two ends of the spectrum of support that shoppers need – and do their jobs as well as the experts could.

Brands can level up with the best in the business in how they craft product description pages. Combined with the ability of platforms like Google to finely understand what converts well, language-model-based AI tools can even orchestrate PDPs that aren't just well-written but also optimized for conversion. My eyes roll several times over, wondering why Google will deprecate its A/B testing tools. An A/B testing tool with human inputs, with a learning loop behind it will make PDP optimizations a trivial job. Maybe Google already knows enough that we will go to a world where A/B testing is not required. I don't believe that will happen but the incentives of product managers in large organizations is a topic I don't want to wade into.

Unlike SEO where Google has an interest in not allowing search results to be flooded with AI-written content, for advertising, the better the copy is the better it is for Google. Generative AI tools that create good copy, great product visuals, and human-like influencer models are all good news. Unlike search which needs signals of relevance for ranking, ads are one-to-one conversations between the audience and the brand and it's the brand's job to convey trust. Google is not the curator of trust here. So Google will encourage generative AI tools and their outputs to drive ad performance.

So what do brands make of the trend of generative AI? If every brand has the same advantage, who wins? Is it the one with the ability to better prompt engineering? I see some obvious takes on Twitter. They all have the "AI-won't-replace-you-but-the-person-who-uses-AI-better-will-replace-you" flavor. I think it is temporary. Prompting is a hack to solve inadequacies in the models. Progressively models will get better. Prompting will get subsumed into the UI and obscured away over time.

Until that happens, for a narrow period of say half a decade, we will have a sister industry to SEO called 'Prompt Optimization.' Clever copy jockeys will tell you what prompts will work better. Generative tools that are part of big platforms that see the connection between content and conversion will start doing keyword weighting. Just like how we have keyword volume and competition for keywords, we will have a grading of prompts – all of these will go on until prompting becomes irrelevant.

Just like how in spite of the efforts of robber barons like Jay Gould, railroads fell the wayside, unbridled enthusiasm for prompting gains will be a thing for some time, until the human-ingenuity that drives prompt engineering becomes irrelevant. Highways replaced railroads. Will AGI replace the current interfaces for AI?

It's exciting times with Dall-E, Vall-E, ChatGPT, and the myriad set of tools that are on our way. Advertising performance is going to improve. PDPs will not be insufferable. Beyond that, how do we know where the edge for the marketer is going to come from?

Maybe, the answer lies in the product and customer support and not marketing!

‍