What you see here is the last week’s worth of links and quips I have shared on LinkedIn, from Monday through Sunday.
For now I’ll post the notes as they appeared on LinkedIn, including hashtags and sentence fragments. Over time I might expand on these thoughts as they land here on my blog.
Is your company starting to build AI solutions? Try this to save money, time, and effort: think “question marks,” not “periods.”
No, seriously. That’s it.
(Yes, I’ve used the punctuation analogy elsewhere. And that still holds. But this is different.)
Why question marks? Because AI is inherently probabilistic. Everything is a “maybe.” You can hire the most talented people, come up with a catchy slogan, and stomp your feet. It doesn’t matter. You can’t guarantee a win with AI just by saying that it’s going to happen.
When you talk about AI with periods:
- “We will predict customer churn.”
- “We will detect transaction fraud.”
- “We will transform this company by using AI.”
… that way lies dragons. And tears. Because you’ve just committed yourself to an outcome that, through no one’s fault, may not happen.
Try to frame your AI ambitions as questions, instead:
- “How could we predict customer churn?”
- “What if we could detect transaction fraud?”
- “What are some ways we could transform this company by using AI?”
To each question, you can say: “Let’s try to find out.”
You’ve now put yourself, and your company, in the frame of mind that you’re testing things. You’ll know that everything you’re doing is an experiment. That means you’ll allocate time and budget accordingly.
And you won’t have to backtrack later if things don’t pan out.
For all the talk about how (generative) AI will impact jobs … it seems especially well-placed to shift the adtech field:
This is an article about online retailers charging for returns. And there’s a useful data angle in here, too:
“The Free-Returns Party Is Over” (The Atlantic)
It’s in this excerpt:
Amazon has begun flagging items with unusually high return rates so that shoppers know to be wary, which pushes the third-party sellers that create most of the site’s product listings to ensure that they’re accurate.
AI is very powerful, yes. But let’s not forget the power of … simple counting, with a side of summary statistics. This costs Amazon next to nothing to calculate but it can have a material impact on the business.
As a customer, you already check star ratings and written reviews on Amazon. A note to warn you that this item is returned a lot, that’s additional info you can fold into your purchase decision.
(Sellers can still game this metric by buying-and-returning competitors’ merchandise… but the only way for them to game their own return rate is to buy their own merchandise and not return it… Which gets expensive. Much more expensive than paying for fake reviews.)
Metrics, metrics, everywhere. You can’t score high marks on all of them, so you have to figure out which ones matter to you and your company.
Consider this funny and informative YouTube clip from Polymatter:
and then, think about:
**1/ Choice of metrics: **It’s clear that The Cheesecake Factory aims to optimize for each location’s revenue, rather than the number of locations.
2/ How the chosen metrics influence the risk profile: The tradeoff for that impressive revenue-per-location is that CF has fewer locations than other chain restaurants. I wouldn’t go as far as to say they face a strong concentration risk … but one could argue that they can afford to lose fewer stores than most chains.
(Which, again, is a business decision. A wise person once explained to me that risk management boils down to answering “what are you worried about?” and “what are you going to do about it?” Each company will have different answers to those questions, so they will understandably take different paths.)
In case you wanted examples of generative AI that weren’t “here’s how ChatGPT will do my job” …
“Forget ChatGPT. These Are the Best AI-Powered Apps.” (WSJ)
"AI is better than people, warns Octopus Energy boss Greg Jackson" (The Times UK)
Jackson said the success of AI was “unlikely” to lead to job cuts at Octopus as the company was growing rapidly. Customer service staff are able to process emails such as queries about bills far more quickly by using AI to draft their responses, which they approve before sending. This frees them to spend more time answering the phone.
1/ This is the benefit of machine-based automation: it frees up people to handle other tasks.
2/ You’ll notice that the customer service team still checks the output before it goes out to customers. (Remember the rule: Never let the models run unattended.)
Using generative AI to vary ad copy and images for different audiences:
“AI machines aren’t ‘hallucinating’. But their makers are” (The Guardian)