Watch videos with subtitles in your language, upload your videos, create your own subtitles! Click here to learn more on "how to Dotsub"


0 (0 Likes / 0 Dislikes)
[AI Business School] [How have you approached identifying and prioritizing] [use cases for using AI in your organization?] >> Chris: When it comes to AI, I think it's interesting for you to apply it to existing business problems that you need help with. And so rather than starting with some random set of ideas that have nothing to do with what you're already doing, for me, it's always interesting to look at the most important processes that you already have as part of your function and then think about how to apply AI to that problem set. We've tried to get very creative and apply it to lots and lots of different things, but I think it's important to be rooted in the real business problems or challenges that you're trying to solve, and then figure out which ones AI can be applied to effectively. [How have you ensured that marketing employees] [have a voice in applying AI to their work?] The interesting thing about AI is it's sort of useless all by itself. You need to have people who are experts in their particular area to figure out how to then apply AI, and the data scientists who understand AI, to the actual business problem. So you can't really do it in a vacuum. You need the marketing expert who understands how to think about content, for an example, that they might want to send out on their different channels, and then how can AI help them figure out which content is going to be more effective ahead of time. So it can't just be done with a bunch of engineers or a bunch of data scientists all by themselves. You need the sellers or the marketers or the finance people who are very deep in their particular business processes to work hand in hand with engineering, with data science, to figure out how AI can be applied. [What advice would you give to other business leaders] [as they determine] where to apply AI] [inside their business?] When it comes to applying AI, I think the immediate notion is to get all excited about the AI and then look around for what do we do with this? I would flip that on its head and I would say make a list of the ten biggest business problems that you have or the ten most important that you have. Those are different for sales than they are for marketing, than they are for finance, than they are for legal, than they are for HR. But if you make that ten most important processes that you have in your function, then you can say are any of these well-suited to big data? Are any of these well-suited to machine learning to applying techniques to make that business process better. Our finance team picked our revenue forecasting, an unbelieveably important business process to the health of Microsoft, and they did a lot of amazing work to build AI models to take a lot of the guesswork and a lot of the seller work out of predicting what our revenue would be. That's a great example of a very important business process for any company. For marketing, content optimization has been what I've been very excited about. Lead scoring is another one I've been very excited about. These are problems or processes that are important regardless of the tack. You have to get better at these as a marketing department then you say okay, great, is this well-suited to AI actually making us much, much better? And that becomes pretty clear you need massive amounts of data, you need to apply machine learning to it, and if those problems are suited to that, then you've got a really great example of something you can apply AI to. [Were any cultural shifts] [necessary to make AI a success?] Certainly I think that AI is here to amplify what human beings already do incredibly well. Getting people to trust the models that we built, once the model becomes very good, you don't need your marketing team or your finance team to worry about what the model is telling them. The model is going to be far better than what they could do prior because it can reason over billions of data points. Getting people to understand that you can trust this model and now you can spend your time on higher value things. You don't have to worry about, is this piece of content better than that piece of content? Feed that into the model, the model will give you the answer. Instead, you can worry about what you want that customer to do, what that next step is that you want to engage with that customer, how is that going to play out. Getting people to move beyond what they used to do and trust the model for it, that's amplifying what they already know how to do. Now you can move on to higher value tasks. There's no shortage or work to be done, and I think getting people to realize, to trust the model and move on to more important things, that takes a little bit of time, and that's something we're still working on. [Did you encounter any areas areas where AI fell short,] [and if so, how were these addressed?] AI fell short where you have small data sets. You need big, big data sets for machine learning to do a fantastic job because there are typically so many variables that you need to take into account. So the first thing you really need to make sure is you're applying it to something with large, large volumes of data. In our case, that's typically petabytes of data. It doesn't have to be that big, but I would say make sure you're thinking about a big data set. The other thing, of course, is that the models have to have deep understanding of the behavior you're trying to drive. So if you're doing lead scoring, you have to make sure that the information you're providing into the engine, or the model, takes into account what a good lead looks like, and doesn't get confused by a bad lead. If you have someone asking for customer support on your website, that's not likely going to be a good lead. You need to weed out understanding what the person is asking for and making sure you're applying a model to the right set of data. [Has using AI created opportunities] [for collaboration between different organizations?] One of the really powerful things about AI is you're combining engineering with deep subject matter expertise in different functions. So that means that the partnership between, let's say, marketing and engineering is going to get much stronger as you try to figure out how to do content optimization. Applying AI to the content that you put in your different marketing channels, that brings marketing closer to engineering, and that means, at least in our case, our products got better in terms of allowing us to fly content through our products themselves. Like the Windows lock screen, where we show an image on that lock screen to hundreds of millions of people. We were able to have marketing partner with engineering to build a solution that allows us to us AI to figure out ahead of time which images would be more pleasing to our customers. The same thing you could say for sales. As we apply AI to our lead scoring, we have to partner really closely with the sellers to make sure we understand what they're looking for in a lead so that we're doing a great job scoring that lead with AI and that brings marketing closer to sales. The same thing would apply to finance. As we apply AI to our revenue forecasting, that brings the marketing teams closer to the finance teams as they look to build those models together. So it's been a great opportunity for better collaboration across functions inside of Microsoft. [Microsoft]

Video Details

Duration: 7 minutes and 35 seconds
Language: English
License: Dotsub - Standard License
Genre: None
Views: 7
Posted by: csintl on Jun 18, 2019


Caption and Translate

    Sign In/Register for Dotsub above to caption this video.