We asked Jonas Dahlberg to give you some insight into who he is, what he does at Microsoft and some important information about the industry.
“What is it you do for Microsoft in your role working with Sales Manager Data and Artificial Intelligence?”
“I work with optimizing the value of our Azure cloud service, which includes AI services. When it comes to getting the value out of AI, access to large amounts of critical data needs to be improved through IT modernization. Getting into the cloud, in whole or in part, is relatively easy today and has great advantages.
Because we at Microsoft offer large-scale data center operation with the very latest technology, our solution is more efficient than if every organization should build, create, and then dimension these environments themselves. The cloud today is not only safer it is also economically advantageous. The cloud allows customers to scale up and down the use of IT and simply pay for what they use.
Many of our customers store large amounts of data in our cloud and use AI and machine learning to optimize their operations. Machine learning requires a lot of knowledge so we also offer ready-made AI services and machine learning models in the cloud, so-called cognitive services, which are easy for, for example, a developer to use to give his application the ability to recognize images for example. Within the cognitive services, we offer support for languages, speech, images, video, text-analysis and much more.
An exciting example right now is the Swedish construction industry that works with Workplace safety. In these environments, cameras are set up in the workplace, which in turn are “trained” to detect deviations. For example, the AI can distinguish whether a person is wearing a helmet or not, if a worker is standing under the crane truck (which is not allowed). Then the supervisor can get the information in real-time and fix the situation accordingly.
What can be a problem is where computing is taking place, in the camera itself or the cloud. The customer may have this feature connected to the cloud, but it often involves large costs to send the data that the camera creates in the cloud. So you create the AI model in the cloud with all the data needed to then create the model to respond to what you want it to react to and then let it run in the camera itself.”
“What would you say everyone is talking about when it comes to AI right now?”
“You can say that there are 3 areas or points that are most talked about right now.
No. 1 is the technology itself.
How does it work? What is it doing? What can you do?
Although the word itself flourishes high on the topic of conversation topics, there are surprisingly few people who know what AI can do and contribute.
What technology will workplaces need in the future and how will they implement AI functionality and get the data needed to train the AI?
When I was talking to a senior executive about AI and he attempted to describe it, it was clear that he not only does not know what AI is, he has no idea what the purpose is and how it can be beneficial. So, I have made it my mission to inform people so that they get a grip on the technology and the opportunities with it.
No. 2 is the ethics.
What happens when models (AI) are trained to make decisions and these turn out to be based on questionable data?
Who should take responsibility for the model acting in accordance with the company’s ethical guidelines?
Microsoft as a company does a lot to ensure that AI contributes positively to the world, see for example the initiative “AI for Good”. Another example is that we publish ethical principles for our AI and have also put in place a monitoring function within our organization.
No. 3 is the value of AI
The companies that have understood the technology and come further into use (found models that work) have continued to find it difficult to get the value out of their business. Often these problems are based on having old technology for data storage and applications in place which limits the ability to distribute the results of the model. Creating a “shopping recommendation” model for a customer who enters a store based on age, gender, purchase pattern, etc. is not that difficult.
But then when this recommendation is to be served to the customer in a store. How is the recommendation then communicated? How can the person (who is obviously going to wear blue clothes with a certain size) receive that recommendation? Via a screen? Via audio in headphones? By the glowing hangers indicating what they should buy?
There is still a lot to do here. The examples we see where AI delivers the most value are where our customers optimizes existing processes using AI.
“So what will you offer us during the Analytics Day?”
I will discuss business cases and report on how AI works. I will show you a couple of different demos on what you can do with AI today.
Thanks Jonas! We look forward to meeting you in two weeks on May 7 on the Analytics Day!