• Stop being a LURKER - join our dealer community and get involved. Sign up and start a conversation.
AI definition aside, (which is a great point regarding definitions @ryan.leslie) let's assume the AI and ML the advertising companies possess is true/reliable and also capable of extracting valuable information that would otherwise go unnoticed. AI and ML could provide remarkable insight for dealers!

In addition to the challenge I think @jon.berna was highlighting in great "database architecture" (buzzword 2022 prediction) is that AI can’t distinguish between good data and bad data on its own, and the algorithms powering AI must assume the data being analyzed is reliable and clean. Bad data, at best, will produce results that aren’t actionable or insightful. Bad data can lead to results that are misleading. In addition to the time and money wasted analyzing bad data, AI can encourage a company to take steps that are even more wasteful. With many dealers having a glut of data, the bigger question becomes: How is the quality of the data?

How many dealers make it a priority to "clean" their data regularly? Prediction: Very Few. Even simple things like NCOA scrubs, info appends, and removing duplicates aren't common practice in automotive.

***Disclaimer: I used AI and ML to make the prediction above.
 
Hi,

Thanks for your thoughts. I think I need to clarify my position. Currently, I see a lot of marketing for vendors in the auto space using AI.
I get a little annoyed because when I think of AI I think of " Terminator ", my mentions of Google and IBM previously. That could be classified as my issue. You are correct that currently you can basically call any SaaS offering AI based upon your Webster's reference.
To Jon Berna's point above.I would love to know more about how Car Story uses AI in their products.
In reference to the gymnast joke. I didn't say I was qualified to teach gymnastics. I said I could call myself a gymnast.

Hey @JayKelly,

I appreciate the discussion. I believe that you and I fundamentally have the same issue. Throwing "AI" into your marketing because your application meets Webster's standard and hoping no one asks what you mean is not beneficial to the dealer community. That frustrates me too.

AI can’t distinguish between good data and bad data on its own, and the algorithms powering AI must assume the data being analyzed is reliable and clean. Bad data, at best, will produce results that aren’t actionable or insightful. Bad data can lead to results that are misleading. In addition to the time and money wasted analyzing bad data, AI can encourage a company to take steps that are even more wasteful. With many dealers having a glut of data, the bigger question becomes: How is the quality of the data?

@Zhendrix,

EXCELLENT POINT! The algorithm isn't as important as the quality of the data.

I understand things through anecdotal reference and analogy. That is just how my brain works. I'm sure that there are others like me here in the forums. I think this is a terrific historical and anecdotal reference to illustrate the point.

The Legend of Abraham Wald

The year is 1943. American Bombers are suffering heavy losses to the German Air defense. The military command solicits the Statistical Research Group at Columbia University to determine where they needed to armor their aircraft to ensure they come back home. They ran an analysis of where planes had been shot after bombing runs, counting all of the anti-aircraft strikes, and plotted the results.

Screen Shot 2019-06-18 at 9.39.22 AM.png

After collecting and plotting the data, it was obvious to see the places that needed to be up-armored were the wingtips, the central body, and the elevators. That’s where the planes were all being shot.

Abraham Wald, a statistician, disagreed. He thought they should better armor the nose area, engines, and mid-body. Which was crazy, of course. That’s not where the planes that were returning were getting shot.

Mr. Wald realized what the others didn’t. The planes were getting shot there too, but those planes weren’t making it home.

Screen Shot 2019-06-18 at 9.55.48 AM.png

What the team thought it had done was analyze where aircraft were suffering the most damage. What they had actually done was analyze where aircraft could suffer the most damage without catastrophic failure. The planes that had not returned had been shot in those areas and crashed. Without Wald, they would have armored exactly the wrong parts of the plane. The team from Columbia weren’t looking at the whole sample set of data, only the survivors.

This is an example of what is known as "survivorship bias." This is a technical term for what we all know well: the dead don't often get to tell their side of the story, and yet sometimes it would be better if they did. In this case, not considering the lost craft is the source of all kinds of misinformation.

The scope of survivorship bias is greater than AI. This is purely rhetorical but interesting to think about. What "lost craft" have we failed to consider? How many current sales processes have been built around the sales that were made and completely ignore the sales that were lost?

-If you sold a car once by calling a prospect every hour on the hour until they bought, do you know how many you didn't sell because of that approach?

-If you refuse to give pricing online or over the phone because your perceived closing rate is higher when you "just get them in" first, do you know how many potential customers never sent a lead?

-If you were successful once overcoming a bait and switch accusation or avoiding a "we owe" due to an error in the listing, do you know how many customers you've lost due to those same errors?
 
AI circus, mid 2019 update – Piekniewski's blog
https://blog.piekniewski.info/2019/05/30/ai-circus-mid-2019-update/

So there you go, the state of AI in mid 2019. As I expected the scene is becoming more and more absurd, essentially a clown show. I think we will soon learn that all that AI bubble was indeed just a joke, something I've been saying in this blog from the very beginning.
 
Almost... every 5 years or so, product managers run out of ideas and decide to take another stab at attribution. I would continue your timeline:

2019: A.I.
2018: Digital Retailing
2017: Attribution
2012: Attribution
2007: Attribution
2002: Attribution
1997: Attribution

Look for attribution's next round of start-ups coming to an NADA near you in 2022, featuring, of course, A.I. :tiphat:

I think you could fit "Responsive" somewhere.
 
  • Like
Reactions: john.quinn