Why We’re Building AI Without a Playbook: Microsoft's Ashley Tarver on Ethics and Responsibility

Summary

Podcast Summary: Liftoff with Keith Newman

Episode: Why We’re Building AI Without a Playbook: Microsoft's Ashley Tarver on Ethics and Responsibility
Release Date: June 25, 2025

---

Introduction

In this insightful episode of Liftoff with Keith Newman, host Keith Newman engages in a compelling dialogue with Ashley Tarver, Microsoft's Evangelist of Data and AI. The conversation delves into the complexities of building artificial intelligence (AI) without a predefined roadmap, emphasizing the ethical and societal responsibilities that accompany technological advancements.

AI Infrastructure and Its Purpose

The discussion begins with Ashley Tarver shedding light on the foundational aspects of AI development. She emphasizes the importance of building robust AI infrastructure to enable effective AI deployment.

“Today's event's been focused mostly on AI infrastructure, how to build the fundamental elements of the infrastructure to enable AI.”
— Ashley Tarver [00:20]

Tarver highlights that while the technical groundwork is essential, the true value of AI lies in understanding its long-term benefits for humanity. She underscores the necessity of envisioning the ultimate goals of AI beyond its immediate applications.

The AI Paradox

A central theme of the conversation is the "AI paradox," a concept introduced by Tarver to describe the fundamental disconnect between human and AI communication.

“Humans and AI don't speak the same language. We're alphabetic and they're binary.”
— Ashley Tarver [00:33]

Tarver explains that humans are now tasked with educating and training AI systems to align with our vision for the future, despite the inherent differences in how humans and AI process information. This paradox presents a significant challenge in ensuring that AI development aligns with human values and objectives.

Ethical and Societal Considerations

Keith Newman steers the conversation towards the ethical implications of AI, prompting Tarver to discuss Microsoft's proactive stance on AI ethics and security.

“Most companies, especially Microsoft, were very diligent about AI, ethics, security, and those core elements to help make this a successful journey.”
— Ashley Tarver [01:49]

However, Tarver acknowledges the difficulty humans face in conceptualizing a utopian future devoid of societal issues like war, crime, and famine. This makes it challenging to program AI with a clear vision of an ideal future.

“We have never really experienced a nirvana scenario... it's hard for us to really visualize what a perfect world looks like.”
— Ashley Tarver [01:49]

Practical Usage and Examples

When prompted about effective AI applications, Tarver provides tangible examples of AI's current positive impacts.

“There's a lot of convenience use cases... the service industry is a perfect example where service tickets being able to answer service problems.”
— Ashley Tarver [02:40]

She illustrates how AI can streamline operations by resolving service issues more efficiently, thereby reducing redundant tasks and enhancing productivity across various industries.

Collaboration and Organizational Strategies

The conversation transitions to the importance of organizational collaboration in AI development. Tarver advocates for the establishment of Centers of Excellence (CoE) within companies to foster multidisciplinary collaboration.

“If you're in a company of any size, you should build a center of excellence on AI... bring in people that will challenge what you're doing.”
— Ashley Tarver [04:17]

By assembling diverse teams, including external advisors, companies can ensure that multiple perspectives are considered, promoting ethical and strategic use of AI. This approach not only unites different departments but also gives everyone a voice in AI-related decision-making.

Addressing Skepticism Towards AI

Acknowledging the skepticism surrounding AI implementation, Tarver draws parallels with historical technological advancements, emphasizing the dual-edged nature of innovation.

“From the steam engine to nuclear power to the Internet, there's been a lot of positive attributes... but they also brought negative consequences.”
— Ashley Tarver [05:13]

She underscores the necessity of balancing AI's benefits with its potential risks, advocating for responsible development practices that mitigate negative impacts while harnessing positive outcomes.

Measuring Success and Future Outlook

When discussing metrics for AI success, Tarver admits that definitive measures are still evolving. She anticipates that as AI technology progresses, regulatory frameworks will emerge to guide and define ethical standards.

“Regulation is kind of the voice of the people... it's hard to tell where that will go. But ultimately that will be necessary.”
— Ashley Tarver [06:27]

Tarver envisions a future where legislation plays a pivotal role in shaping AI's trajectory, ensuring that its development aligns with societal values and ethical norms.

Conclusion

The episode concludes with Tarver and Newman reflecting on the delicate balance between advancing AI technology and maintaining ethical integrity.

“As we build the technology up, how do we make sure we're building it in the right way?”
— Ashley Tarver [05:49]

Tarver emphasizes the collective responsibility of developers, companies, and regulators to steer AI development towards beneficial outcomes while safeguarding against its potential pitfalls.

---

Final Thoughts

Ashley Tarver's insights provide a nuanced perspective on the challenges and responsibilities inherent in AI development. Her emphasis on ethical considerations, collaborative strategies, and proactive regulation underscores the multifaceted approach required to navigate the evolving AI landscape responsibly. For listeners interested in the intersection of technology, ethics, and societal impact, this episode offers a thought-provoking exploration of building AI without a playbook.

---

Listen to the Episode: Liftoff with Keith Newman on Apple Podcasts

Transcript

[A]

All right, we just got off stage. Ashley Tarver, the evangelist of data and AI at Microsoft. Evangelist of data and AI at Microsoft. That is a big title.

[B]

Thank you. Well, yeah, it's vague enough that nobody really knows what I do, so it's perfect.

[A]

Yeah, I'd love to have that title myself. So how did the talk go? We were talking about the AI paradox a little bit and a little bit off camera. We were discussing that. Share with the audience here what you mean by that and how you look at that challenge.

[B]

Yeah. So today's event's been focused mostly on AI infrastructure, how to build the fundamental elements of the infrastructure to enable AI. So I took the audience on a break and give them a chance to think about the future and what the AI is really intended to provide to the humanity in the long term. So as we're doing our tactical work to get the AI built and deployed, the real value is understanding why we're doing it and what it will ultimately allow us to do into the future. And with that, there's a massive paradox because humans and AI don't speak the same language. We're alphabetic and they're binary. And so we're learning with these chat engines today how humans can better communicate with the AI. But at the end of the day, the paradox is the fact that we're expected as humans to educate and train the AI on what we think our future should look like.

[A]

So where does that take us? Where do we start to engage and not just drive it from a technological perspective or an infrastructure perspective, but a societal, ethical, bingo, correct way to build?

[B]

That's exactly right. And we're doing that now. Most companies, especially Microsoft, were very diligent about AI, ethics, security, and those core elements to help make this a successful journey. But the bigger problem is we as humans have never really experienced a nirvana scenario. We've always had war, crime, famine. Famine. These are real life scenarios that we've all know, pandemics. And as a result, it's hard for us to really visualize what a perfect world looks like. So. So in that example, how are we expected to train and program AI to deliver on that future promise?

[A]

Ashley, can you give us an example of how companies are using it correctly? I think there are enough examples probably incorrectly.

[B]

Correct. Yeah, yeah, no, there's plenty of good use cases today. I mean, we're in a very tactical stage of our long term strategy with AI. But if you look at the immediate, there's a lot of convenience use cases, I would call it, where AI is helping Shave off redundant tasks. The service industry is a perfect example where service tickets being able to answer service problems. You generate a ticket because your software's not running properly. The AI can really help resolve that more quickly.

[A]

I'm talking more about the societal or ethical considerations too, and how things are getting deployed and how companies need to pay attention to. To those elements as well.

[B]

Yes, we're not ready there yet, to be honest with you.

[A]

How do we get ready then?

[B]

Yes, that's part of where we are. We're so early in the journey right now. If you're being ethical, you're not really using AI to establish Personas or to identify people, culture issues, because right now it's not really smart enough to do a good job at that. So as a result, we're being, being careful on how we use it in the area of like facial recognition or identifying personal attributes. Those are futuristic opportunities, but right now it's a little more tactical.

[A]

Yeah. And in terms of how you scope your role, which could I imagine be pretty broad. But how do you get people to collaborate more? Right. There's the technology side of the house that's driving really fast and hard. But how do you organizationally build so that these considerations are in the proper place?

[B]

You know, we're actually doing a live workshop this afternoon on that topic. What can companies do tactically in a strategic way? And one of those areas is to. If you're in a company of any size, you should build a center of excellence on AI. Basically, you pull together elements of all the teams into a center of excellence. Tiger team, if you will, where you bring in, don't bring in like minded people, bring in people that will challenge what you're doing. Even outside people into the company to give advice on how you're directing the use of AI toward yourself, your own company and ultimately your customers.

[A]

And that'll bring people together?

[B]

Absolutely.

[A]

Yeah.

[B]

Yeah. It allows everybody to feel like they have a voice at the table.

[A]

And what about the, what about the. You have all the AI advocates, What about those that are a little bit wary of diving in and not implementing it?

[B]

They have every right to be. If you look at the history of our technology evolutions, from the steam engine to nuclear power to the Internet, there's been a lot of positive attributes associated with these technologies, but they also brought negative consequences. And so that's the key right now in our ethical path. Paradox is how do we manage the good and the bad at the same time? And it takes a family.

[A]

Yeah, it does. Interesting. So how do you see things being reshaped not just within Microsoft, within the tech industry, within the AI industry, but also just society.

[B]

It's going to be a joint effort. As the technology gets better, the capabilities will deliver new value that we don't see yet. But concurrently, we as the programmers have to be considerate on how we don't bring our biases and our ethical dilemmas and our own egos into the development of that AI. That's where the paradox is. You know, how do we make sure as we build the technology up, how do we make sure we're building it in the right way?

[A]

Do you have a way of measuring success in terms of this broader goal, in terms of this metric in mind?

[B]

I think the jury is still out on that. But you know what will happen as the technology continues to expand? I think you'll start to see regulation come in. And as a result, regulation is kind of the voice of the people. I mean, arguably our elected officials will help bring our value elements into those regulations. And it's so early right now that it's hard to tell where that will go. But ultimately that will be necessary.

[A]

Ashley, it's been great having you here. You're a great balance to the conversation around infrastructure and technology.

[B]

Thank you.

[A]

But to balance things out in terms of how we are using it.

[B]

Why are we using it?

[A]

Why are we using it?

[B]

Exactly.

[A]

So fantastic. Thanks for coming in. I hope everybody will go check out your talk as well.

[B]

Thank you.