Part 3: Leveraging Azure OpenAI to Grow Your Business

Irina, in our last conversation, we talked about how companies could leverage Azure OpenAI service and other various cloud solutions that Microsoft offers to transform their businesses. Microsoft has such a strong industry focus on cloud. How would an enterprise or ISV that's focused on regulated industries think about using Azure OpenAI or Microsoft's other cloud services?

For an ISV in an industry specific area, it's really important to know where you specialize. There are a ton of capabilities that are industry specific and in need of transformation. For example, one of the key industries undergoing seismic transformations is healthcare. Anyone who's ever been to the doctor or tried to schedule an appointment can attest to this. There are so many opportunities right now, especially with use cases like virtual visits, getting your prescription filled online through a virtual pharmacy. While we're changing, we're not changing enough. It's still a very broken system. On the other hand, what you have is a highly regulated industry. Changing anything in this industry will be tough. And especially when you come in with AI solutions and AI cases, it will be tough, especially when it comes to the patient care model and patient support and providing advice to doctors driven from a variety of inputs, including medical devices that analyze patient signals. That will be super tough nut to crack and we will continue to work on that with ai solutions.

But there are a ton of areas where I think we will see a huge change happening from AI that will improve productivity, reduce cost and drive more personalized customer care.

• We will see a lot of opportunities in the back-office management. So, claim processing cost estimation etc. actually can be automated. There are a ton of things already happening there. Leveraging Chat GPT with semantic search, you will see that that processes can be automated, they can be much faster, and you can provide much faster understanding in interactions with the patient, along with right cost estimation.

• If you've ever worked with healthcare, cost estimation is really challenging as there are a lot of variables at play depending on which system you are in, who your insurer is, etc. There's an opportunity to democratize cost estimation leveraging AI.

• An easier point of entry is really enhancing patient engagement. If you've ever been in any hospital, you know that how many forms you need to fill out . Much of this can be done at home, reducing the need to scan forms and so forth. There is a ton of opportunity there.

• Hyper personalization is another opportunity – right now, patients are treated as one big group and analyses and treatments are based on which group most closely resembles you. But a lot of things from medical perspective can be hyper personalized because we have so much data on that person. Our fitness trackers let our doctors know how much you exercise, they know what medications you take and these things can be used and incentivized to create much more holistic view of the patient that allows you to treat the patient better.

We have the opportunity to scale to our customers and partners to improve clinical and operational user insight. Interactions with a patient does not stop after the doctor sees you. There is a lot of interaction has especially with the long term care or is the chronic care. You need to think about how to constantly interact with the patient and help them to get better. How can you incentivize them to be more active or lose weight? Or take actions to have more positive impact on the patient in the treatment program?

Financial Services and Insurance is another traditional and highly regulated industry, but there are also opportunities there, especially for a Microsoft partner. For example, if you're talking about contact center analytics, there are specialized services around call center, insurance, banking, capital markets, how you do financial advisory, how you can help clients when you interact with them and make the interactions with customers much more positive, how can we upsell or cross-sell opportunities as you're interacting with the customer. Regulatory compliance is interesting as well. For example, in compliance industry, how can you build a solution to help agents with underwriting rules and so forth: how can that process be automated and reviewed in such a way that it's in compliance with regulations in this environment. I think it's going to be great to go deeper in and explore each of these industries, talk about specific cases and do some brainstorming with partners and ISVs about pretty exciting opportunities that we have today.

Yes, I agree. We should do deep dives into the opportunities for people to get creative in optimizing solutions for regulated industries. There are two things I'd like to ask you. One is, innovation. How do people get this innovation started? The second is, people are terrified that AI will replace humans in the workplace. But it sounds like from what you're saying we will be able to coexist. What are your thoughts?

Let's unpack a few things. First of all, let's talk about replacement of humans. I think every sci fi that I read since I was eight years old talked about the AI apocalypse. AI is like a savant type of thing. It's the intelligence of a very, very bright kid with the language and social skills of a three year old. It could be spitting out a lot of information, but that doesn't necessarily mean it comprehends what it is saying. And I think it's really important to understand that AI in the state it is in today will augment and support, rather than replace, humans.

I think there is a lot of fear. For example, people are worried that creative writing will be destroyed, but you need a prompt that actually creates something creative, and requires the person prompting the machine to think through things, it requires the person to put ideas together and the input needs to be a certain combination of things, and all of this is still human. It's not created by Chat GPT.

Will it be possible at my company at some point to do this? Probably but at this point, we're not there yet. I think that people are concerned about how creativity can be destroyed using DALL-E but I think it's actually the opposite. As a person that loves art but cannot produce art, I will say that for me, it opens up opportunities where I can potentially construct something visually that I can later show my company or the artist and say this is what my inspiration is right now. This can transform even simple things like creating company logos. As a company that might not have money or resources to pay great artists to craft your company logo, this is maybe the first step for you to try to create something that is yours and is simple, that artists and professionals can bring to life.

I don't think it's going to be a replacement. It's going to be augmentation. Technically, some of the AI ability can better detect cancer that humans have a hard time to detect by looking at images. Again, I don't think anybody right now will just trust AI and not ask a doctor to review an image or this type of image side of thing. So I will not say that this is coming and the image of jobs will be replaced.

That being said. I think responsible AI is super important. Understanding that AI can be used for good or for bad is really critical today. Solutions don't have a moral compass. If asked to do something malicious, AI will not detect and say no, no, you can do this. There are certain barriers that have to be created and some that have already been created to ensure that we're not crossing this line, but if you follow the news, we can see that some of those barriers can be broken. Microsoft has been investing in Responsible AI for over seven years now. We have specialized teams that work on that and ensuring that we are reviewing things case by case.

One of the challenges right now, according to a lot of partners, is access to Azure OpenAI. The reason why partners say it's so hard to access Azure OpenAI service is because there is a human review happening to understand who is trying to use this, what's your use case and how you plan to use Open AI services to ensure that we at least try to protect the general population and protect our society from malicious usage of OpenAI. It's such a new thing that I think we will have to live through this for a while, to establish the rules and establish regulations to help understand what kind of potential new laws needs to be put in place. But it's an AI explosion right now. And we have to think about this. How do we maintain this while we're still exploring opportunities, to explore the capacity and power of these new technologies?

It's like evolution - unless you manage and put guardrails in place, things can just go haywire. So that's a big business challenge for us to address.

It's going to be a big problem. There is already malicious code ready capable of creating viruses. While it's not created using OpenAI, the technology is there and the capabilities are out there, leveraging the engineering resources and power and models that exist. These generative models use the resources and a massive amount of data. Today, infrastructure should be cheap. If you really want to train the models and spin out computer viruses, it's not that complicated, actually, as scary as this sounds, the technology is there. But then you have people who try to protect us using exactly same resources and the same security tools to defend against that.

These tools can be leveraged for good, there are ways to identify new drug formulations that can be used to treat diseases. However, we need doctors and medical professionals to review the data and decide which treatments should be used on which diseases. The magic of these technologies is that they increase the speed to development of medications. So we have two extremes, the really good and the really bad, and the use of this technology is basically limited by your imagination.

It's incredible and makes you wonder that if it's almost table stakes to learn how to code for this generation, is learning how to program or leverage AI going to be table stakes for the next one?

Yes, I think it's if you're looking at the cloud and the future of technology, there's going to be actually multiple tiers of leveraging AI. It could start with low code. Today at Microsoft, we have AI builder, chatbots, Power Apps infused with AI technology that you can use to build today fairly easily. That being said, you have to have a common sense to understand how do we validate that what the model was spitting out is correct. How do you evaluate that and this is where mathematicians come in due to statistical understanding of how this model works, and this has become really difficult because a lot of these models are black boxes, or have quite complicated mathematics underlying them. Moreover, advancements like GPT-3.5 Turbo or GPT 4 contribute to the evolving landscape, introducing new dimensions to the challenges and opportunities within AI.

The question is really becoming how do I understand what to trust and how do I put checkpoints around things to ensure that what I do with this is not dangerous and put anybody in danger. But you don't necessarily need to build AI models. There are a lot of companies today that actually focus on creating these models pretty efficiently. The jobs of the future will have lots of people who have mathematical background, you know, Revenge of the of the old nerds. It'll be crucial to have a deep understanding of the models we build.

You are going to have access to a lot of people with different jobs. And I think this is going to be a new generation of developers that actually focuses on the AI only and it's important to understand that but as you said, this is democratization. There is going to be a different level of access to training and skill sets and different opportunities for each of the skill sets that will be available there. But not knowing AI is probably not going to be an option if you want to really have access to a good job in the next 10 years.

Do you have any last tips for people looking to get started in the AI ecosystem?

There are great tools today that are available, like learning.microsoft.com where you can go and actually learn a lot of things. I highly suggest that you actually get yourself familiar with that and at least understand the Microsoft partner ecosystem. I think it's important for the partners to go on partner.microsoft.com and understand what solutions and capabilities are available there. Again, to understand how to take advantage of the high level solution to build faster, to go to market faster, and how Microsoft can enable them to do that because there are tons of problems right now available for partners to get engaged in Microsoft community. There is huge interest to bring a lot of the benefits to partners in this space. It's going to be really important for partner to spend some time learning, practicing and understanding the social space because it's pretty big one.

Want to learn more about Irina Petrakova-Otto? Follow her on LinkedIn! 

Want to learn more about leveraging Microsoft’s ecosystem and Azure OpenAI? Follow us on Partner1 where we will be sharing insights from leaders like Irina on how companies can leverage AI to grow their business, and more.