Ancient Grains & Quality Info
The NY Times reports 8 incidents on United of late and right about the same time its Premier members receive an email about safety from UA’s CEO Scott Kirby. This is not the type of email any airline should ever have to send.
Imagine:
Dear Loyal Banking Customers: We’re writing to let you know there’s been a few minor hold-ups in our branches of late but rest-assured your money is safe and we run a culture of security in all banking operations….
This doesn’t make anyone feel any better. Because when you fly on planes or deposit your money in banks you are not thinking about safety or whether your money will vanish. It’s presumed. Quality is built in. It’s table-stakes for keeping your doors open, your lights on and your planes in the air.
And it doesn’t make us feel any better when we use words like ‘incidents’ or recently with LATAM and Boeing’s Dreamline: technical event — to leave us wondering and then finally tell us it’s human error when a flight attendant bumps into the wrong button and sends the pilot toward instruments that basically disable the plane. Sorry folks — none of this is settling. Whether human error, software glitch or, or, or. We don’t care. Because no defense or explanation is defensible.
Imagine when we start getting these messages about our AI algorithms. We’re sorry we let the wrong people out of jail because there was bias built into the algo and by the way there were no humans in the loop.
We hate to tell you that the logic for whether the traffic lights worked, your self-driving car crashed, or worse the reason it hit someone was because the data we used to train the LLM was bad, incomplete, or worse we didn’t even know what data was used to train the dang thing.
We’ve said before and we’ll say again data in our algorithms is like the ingredients in our food. They are listed in ascending order by importance. The exact recipe isn’t given out, but the proportions are. At best we’d love to know the source (organic farms provided those eggs and ancient grains and here are their names names!)
For algos it’s going to have to get to that level or we’ll never have trust in the algorithms nor in the bias they exercise, the degree of that bias, and how we can eliminate it as best and as much as possible. Machines don’t code themselves. People do. And where there are people there is bias, unconscious, or otherwise. It’s as simple as that.
So, we have to do our best to deliver quality. And that’s how we extract value and ultimately price for our data and information when we license it to others or use it in our own commercial offerings. Sure, it’s about the use case but it’s about prepping it for that use case and understanding where it fits in the list of other ingredients, and its provenance, currency, relevance and dare I say quality.
My colleague David Worlock, Outsell’s Chief Research Fellow, and Industry Futurist, who is much closer to AI, predicts we are going to have to get far better at our data preparation to extract more value from our data and more value also means better pricing.
At the end of the day the algorithm is only going to be as good as its ingredients. And who has the best ingredients when it comes to data and information? You guessed it. Global brands with premium, well-differentiated information. Just like artisan bread made from ancient grains, with seeds and all kind of good grist from the mill — it will extract more value. $12 a loaf anyone? So much better than chemical-laden white bread for $2.99.
