A.I. ain't A.I. in automotive. It is B.S.

[jon.berna]

This topic is in my wheelhouse

I recall a conversation i had last year with a CEO of one of the larger companies in auto. They tell me how committed they are to machine learning in their products because they just hired a data scientist. Oh boy...

I see these roles in a lot of companies now. Most of these companies flat out aren't ready, and these people will leave as soon as they find out the truth.

What is missed by these companies is that 95% of the work in machine learning is in how the data is architected. The value is extracted and compounded after this prerequisite. Database architecture and the development efforts behind it are literally the most valuable part of the infrastructure required to even start considering deploying machine learning. We are talking 1000s of hours here, not a couple months. The machine learning part actually takes a very small amount of time.

Data flow and design is also not sexy, requires insane discipline and gets pushed to the wayside by product teams trying to hit marketing ready user stories. To 99.9% of humans it is boring and impossible to understand.

Full disclosure I love database design, it is truly art, I think it is amazing what can be done to refine raw material into a polished gem. However, I venture to guess most people in senior leadership out there don't and have no desire to learn it. Which is fine, except they dont listen to the engineers that know how to do it right.

Product teams (no offense) are just fancy sales people, they have no business jamming machine learning into a product deliverable.

So to that CEO, instead hire a software architect and a database architect. Support them, give them time, learn from them.

 

[JayKelly]

In my humble opinion, the common ground is a grey area. It is very easy to use the catchphrase "A.I." I also don't believe anyone is truly offering A.I. IBM's Watson is the closest. I have emailed many people in Automotive to caution them about using "A.I " in their marketing.
At Faraday, we use machine learning algorithm's to predict like outcomes. We call it a Data Science layer for your marketing tech stack.

I only read articles about A.I. that are written by someone with a Ph.D. in Data science.
The majority of offerings and articles should just be considered click bait unfortunately.

 

[ryan.leslie]

I also don't believe anyone is truly offering A.I.

I'll repeat what I said earlier for emphasis. You HAVE to define the terms, it is imperative! As an industry, we need to get really good at that, and quickly, or else we are going to do a lot of talking right past each other!

No offense intended at all Jay, but I want to use your statement above to illustrate why the definition of terms matters so much. Saying "I don't believe that anyone is truly offering A.I." may instantly call into question your understanding of the terms and their meaning. I believe you were using the term AI as a substitute for a subset of AI. That position could be defensible depending on the specific discipline, but the statement above is not. Using the broad term kills the conversation. There is a lot of AI in automotive.

AI is defined by Webster's as "a branch of computer science dealing with the simulation of intelligent behavior in computers; the capability of a machine to imitate intelligent human behavior." By that definition, the bar for "what is AI" is as low as automation. AI is absolutely offered in this industry today. Alex's cartoon that started this post depicts actual AI. A machine is "simulating intelligent human behavior" by following a rudimentary, predefined, and binary workflow (this is called encoded knowledge). Is it sexy? No. Is it WestWorld quality, fleshy, sentient beings? Not even close. But is it AI? Yes, it absolutely is.

My opinion only, but vendors need to go a step deeper than "we use AI" in their product descriptions since anything that simulates intelligence qualifies. It's a little like a dealer saying "we sell cars" to a consumer, the follow-up question is "what kind of cars do you sell?" In the same way, when a vendor says "we use AI," a dealer should respond "what subset of AI does your product use?"

 

[JayKelly]

My wife likes to have fires in the fireplace all year round. She like the ambiance and it takes the chill out of colder spring days. I asked her how she thinks the logs feel? The closer you get has a direct corrleation to your opinion.
You are making my point. There is no clear definitve answer.
When I think of " AI " I think of Google's computing divison running for a month to figure out the root of 3.14159265359
IBM Watson commercials for smarter cities.

I dont think of a company that is adding 5 cookies together or is asking a program to follow 6 predetermined steps.
If the bar is set low enough I could be a gymnast.

Jay

 

[jon.berna]

I think there is a clear distinction is between semantics versus fraud.

Two examples:

1. All of the major platforms (Facebook and Google) use AI, that doesn't mean the vendor leveraging those products are. However, they sell themselves as something they are not, aka fraud.

2. When a company truly builds their own platform leveraging a subset of AI it's semantics.

If the answer is #2 the company should be able to show from database to code to visualization layer how they leverage [Insert AI related phrase here] if they can't they don't qualify for #2 they probably qualify for #1.

 

[Alex Snyder]

If the bar is set low enough I could be a gymnast.

I love your way of looking at life Jay.

 

[Alexander Lau]

My wife likes to have fires in the fireplace all year round. She like the ambiance and it takes the chill out of colder spring days. I asked her how she thinks the logs feel? The closer you get has a direct corrleation to your opinion.
You are making my point. There is no clear definitve answer.
When I think of " AI " I think of Google's computing divison running for a month to figure out the root of 3.14159265359
IBM Watson commercials for smarter cities.

I dont think of a company that is adding 5 cookies together or is asking a program to follow 6 predetermined steps.
If the bar is set low enough I could be a gymnast.

Jay

 

[joe.pistell]

 

[ryan.leslie]

When I think of " AI " I think of Google's computing divison running for a month to figure out the root of 3.14159265359
IBM Watson commercials for smarter cities.

I dont think of a company that is adding 5 cookies together or is asking a program to follow 6 predetermined steps.

Jay, this statement, with the added emphasis, is exactly the point. There is a clear, definitive, answer. AI, like every other buzzword this industry has adopted in the last 15 years needs to be defined objectively. Put 10 managers or vendors in a room and ask them to define:

• Online Reputation Management
• Digital Retailing
• Attribution
• Big Data
• CX
• Engagement Metrics

You're going to get 10 different answers. These terms have become ubiquitous, catch-all, junkdrawer terms where everyone has their own definition because the industry allowed that to happen. AI is going to be in the junkdrawer too I'm afraid. That's why dealers should question it every time they hear it and understand just how low the barrier to entry is to claim AI in the application.

Let's start figuring out what these words actually mean instead of assigning our own meanings. It'll make things easier for all of us. AI just means "simulated intelligence by a machine."

If the bar is set low enough I could be a gymnast.

This is a great analogy. The balance bar is low and wide and you can walk across it, so now you're a gymnast. You and I meet at a corner basement bar near the gym because I'm interested in hiring you to train my kids. The first two questions I'll ask you, "what kind of gymnastics do you specialize in? How can I expect for my kid to improve from your gymnastic abilities?" I'm going to learn pretty quickly that you are just a guy in a leotard hanging around a low bar.

Here is the point: There are a lot of companies that accurately claim they have some AI in their applications (I like Jon's descriptions by the way). The bar is super low and wide to accurately make that claim. Rather than saying "that isn't what I thought you meant by AI" we need to be prepared to ask "what subset of AI does your application employ?" The really skilled gymnasts will stand out from the rest.

 

[JayKelly]

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.