Mixing metaphors to build an AI strategy (Part 2 of 2)
My previous post covered half of the four To-Do’s that will set you up for your best AI strategy (based on research that Google and Atos recently commissioned). First, find executive champions to inject an AI-mindset into the company and lead the charge. Second, get your data ready for analytics. Data is the main ingredient of AI. Make sure it’s captured, processed, cleaned and understood appropriately. Then you can focus on finding the best use cases for quick wins.
3. Don’t try to boil the ocean
Thinking small is the secret to big success in AI. Too many companies are trying to boil the ocean and inject AI everywhere, which leads to no success at all.
I was talking with the head of data science at a big company. She said she’s pretty lucky to have a large team of data scientists. So I asked what they’re working on. What are her goals for the team? She said they’re doing too much, actually, and that she’s trying to cut 80 percent of their projects. She is exactly right! The key to AI is to think small.
At Google, we went one application at a time. Google Photo started as a dump storage app. We applied AI to do image recognition, and now you can search your photos by keywords. Then, next project, we looked at our data center energy consumption. We installed sensors everywhere, and now we’re basically letting AI run the data center. Our energy bills went down drastically. With Gmail, we put AI to work for predictive typing. We take one app at a time.
How do you know which projects to take on? Look to your executive champions (from Part 1 of this post) first. It’s part of their jobs. The use cases generally aren’t rocket science. They’re known. The technology exists. The partner ecosystems exist. Be aware of your competition, and listen to your customers and stakeholders. Sometimes the main use cases are right in front of you.
4. Steer the right course
This is about identifying the legal and institutional structures that govern your data. The study we commissioned from IDG Research and CIO Magazine shows an interesting gap between the leaders and followers on this topic. 67 percent of the leaders consider it a barrier, but only 57 percent of followers see it that way. In other words, the people with less AI experience more often underestimate the gravity that accompanies the complexity. The risks of getting it wrong are both operational and ethical.
At Google, we’ve put a lot of thought into this (here’s our blog post about it) in response to a recurring set of topics (worries, really) that our clients raise. Our position is clear: AI programs at the enterprise level in particular must treat people fairly, be transparent, prepare and not replace the workforce, and do good. Data-intensive companies always start off with the best of intentions, then something happens, and those intentions get lost. It’s like the need for dental hygiene. Get your regular checkup, and don’t forget to floss, or you’ll lose you’re your teeth. There aren’t a lot of other options. We know the rules, but for some reason, we don’t stick to them.
Artificial intelligence can do a lot of good things but also very bad things in the hands of the wrong people. We have a responsibility to leave a good world for our children.
It’s the new 90’s of mobile
AI is really, really in the early stages right now. We’re embarking on a journey that will take us through the next 20 years and beyond. Remember the 1990’s when mobile tech was at zero? Now, there’s better than 100 percent penetration. That’s more than 1 phone per person. Everyone uses them for everything.
AI is at that point right now. It’s like the late 90’s of mobile. You have time. But do plan your strategy now, to avoid falling behind your competition. I hope my mix of metaphors helps to guide your thinking: Find a champion, bake the cake before icing it, don’t try to boil the ocean and always steer the right course.
Get full access to the results of the study that Google Cloud and Atos commissioned from IDG Research and CIO Magazine at https://atos.net/en-na/lp/ai-readiness.