Artificial intelligence is revolutionizing project management, including risk management, by significantly enhancing efficiency. Karanveer Anand, a Technical Program Management Leader at Google, offers practical advice on integrating AI into workflows. He emphasizes three key pillars where AI impacts project management, including identifying risks using accurate historical data and ensuring data security.
Chapters
00:00 … Intro
02:16 … Meet Karanveer
03:41 … The Grit to Achieve Your Goals
07:39 … AI And Day-to-Day Responsibilities
08:25 … The Impact of AI on Teams
10:06 … Practical Training on AI
12:05 … Stakeholder Prompts for AI
13:47 … How AI can Identify Risks
16:20 … Accurate Historical Data
19:15 … Data Security
21:14 … Personal Project Experience using AI
23:30 … AI in Resource Management
25:32 … Ren Love’s Projects of the Past
28:15 … AI and the Human Element
31:54 … Addressing Ethical Challenges
33:53 … A Mid-Mortem Assessment
38:18 … Get in Touch with Karan
39:57 … Closing
Intro
WENDY GROUNDS: Welcome to another exciting episode of Manage This, the podcast by project managers for project managers. I’m Wendy Grounds, and in the studio here, is Bill Yates. We’re so glad you’ve joined us; and if you haven’t yet left a review on Apple Podcasts or Google Play, that’s a great way for you to show your appreciation. A quick review of Manage This would mean the world to us. We’re so grateful to everyone who’s already taken the time to do so. Thank you for being part of our community.
Today we’re once again tackling a subject we’ve spoken about before. Bill, do you want to kind of tell us why we’re speaking about AI again?
BILL YATES: Yeah, AI seems to be really important to project managers.
WENDY GROUNDS: Yeah, mm-hmm.
BILL YATES: Artificial intelligence is right at the top of both our interest and our concern as project managers. People want to know, “Hey, am I going to still have a job? If so, what does it look like?” We had a great conversation with Oliver Yarbrough on this topic; and we’re going to go a little bit deeper, I would say today, with Karan. The conversation is going to get into a specific area of project management, risk management, and how we can use AI. Karan also shares some personal uses of AI that really help him.
WENDY GROUNDS: And if you’re wondering why we’re talking to Karan, let me tell you a little bit about him. His name is Karanveer Anand, and he is a technical program management leader at Google. He’s been instrumental in integrating AI tools like Gemini into project management, and he also serves on advisory boards for San Jose State University and UC Irvine. His expertise in leading cross-functional technical programs focuses on enhancing the reliability of Google Workspace services. You’re all familiar with things like Gmail, Docs, Drive, Google Meet, and Google Calendar. We have Karan to thank for a lot…
BILL YATES: That’s right.
WENDY GROUNDS: …of the work that is put into those programs. His data-driven approach has improved the performance of these critical tools. Karan comes from a small town in northern India to a leadership role at Google, and his journey just reflects his hard work, innovation, and leadership. If you’re curious about the future of project management, this episode is packed with some valuable insights and really wise words from Karan.
Meet Karanveer
Hi, Karan, welcome to Manage This.
KARAN ANAND: Well, thank you for having me here. I’m glad I’m part of your podcast.
WENDY GROUNDS: It’s an honor to talk to you. But before we get into our whole conversation on AI and all that we can learn from you and risk management included, can you tell us a bit about your journey? I know you have a very interesting background story, including your career path. How did you end up at Google?
KARAN ANAND: Yeah, I’m a technical program manager in the reliability organization, and I believe there’s no better organization to work for than Google in the reliability space because we trust and rely on Google services so much for our day-to-day lives.
BILL YATES: Mm-hmm.
KARAN ANAND: At least in my circle, we say if my Internet is not working, then we try checking www.google.com. Is Google.com working? It means my Internet is working.
BILL YATES: Yeah, that’s the test. That’s true.
KARAN ANAND: It shows how much we trust in Google.
BILL YATES: Mm-hmm.
KARAN ANAND: I have been in the reliability space, so I thought there is no better company to work for Google. So, I just studied for interviews on how to crack Google, and Google is the pioneer in the AI space, as well. They have been doing deep learning, machine learning for many years, not from last two years, from many years. And they have invented a lot of this transformer architecture. They have been the leader in the AI space for many years. So, I felt Google is the best company to go where I can scale my efforts, I can learn from my peers, and have an impact on billions of people across the globe through technical program management.
The Grit to Achieve Your Goals
BILL YATES: That’s so true. You are so perfectly placed with your skill set, with your background in that organization. And we were really impressed, Wendy and I, as we read about your background. So, I came from a small town in North Georgia. That’s where I grew up and then eventually found my career path. You came from a small town, a small region in India.
KARAN ANAND: Yeah.
BILL YATES: And then went to university and then ended up in California. How did this happen? We know you’re a very bright person. Did somebody just recognize, okay, this guy has talent?
KARAN ANAND: I guess having a strong grit, my colleagues and friends who are having a hard time or a bad time in their life, I think things will get fine as long as you have the strong determination and grit to achieve that goal.
BILL YATES: Yes.
KARAN ANAND: No matter what the problem, Google, USA, Australia, any country you want to be a millionaire, that’s fine. As long as you have a plan, you have a goal, determination, and a grit to achieve this, you will be there one day. That’s a motto of my life, and I’m still following it for my next set of goals in my life and keep growing. But never miss to enjoy life while you’re hustling. So that’s a motto.
Like you said, I came a very small town in India. I stepped out of my hometown at the age of 16 because in small towns there’s always a thing to stay with your family, to not go very out. But I stepped out of my hometown at age of 16 to study in India. Then I started my first job at age of 21 or 22, something like that, in Bangalore, which is Silicon Valley of India.
BILL YATES: Yeah.
KARAN ANAND: I worked there for 18 months, and my company moved me here to America to an extraordinary visa, specialized knowledge visa. They caught me with just 18 months of experience. Out of 18 months, six months took up paperwork. So, my process started after 12 months, and I was very fortunate at that time to go at this opportunity because of my work, my connections, soft skills, hard work, its place, everything has a flavor. Like you make a good food, salt has own flavor, pepper has their own flavor. So, things worked for me. Then I came here, worked here for a company in the reliability field. It was small to mid-sized company. I felt like I know everyone there. I’m a very extrovert personality. I make friends very easily.
I thought being a technical program manager you need to get your work done. And I felt like it’s not challenging anymore because I know people, I can just call them up. “Hey, why are you not working on this? You need to do this. I need this.” Update. It was like a very friendly thing for me. So, I thought it’s not challenging. I’m not growing here. I’m getting in my comfort zone. Whenever a person gets into comfort zone, that’s where the learning gets stopped. So, I just pushed myself.
Then the COVID came, and I thought it a good opportunity for me to learn and grow the new technologies which are coming up. Then I did online master’s while working from home in COVID. In 2020 I started from UT Austin. That’s a prestigious school for computer science. I finished my master’s in two years while working, and I learned a lot of new things. 2020 AI was not too fancy, not too on the top. So, it gave me an edge on how to use this AI LLM techniques.
Then I went like Google is the next place to be. And I cracked the interview and been growing here, getting promoted, seeing what’s next, and challenge yourself for the next thing. Okay, why can’t I do this? Why can’t you do this? And growing. And no matter which field you are in, technical program management is just one field, I’m saying. But AI is at the center of each field.
BILL YATES: Yeah.
KARAN ANAND: So, it’s a pretty interesting space to be in and having an impact to see where the program management goes and how it’s impacting. Because there are not many practical applications, I’m seeing in the market on how AI is impacting the project management or program management. A lot of people are talking about it, but how many people are actually using it?
BILL YATES: Yeah.
KARAN ANAND: And how much is it actually affecting? These are pretty decent things to think about.
AI and Day-to-Day Responsibilities
WENDY GROUNDS: So how do you see AI transforming the day-to-day responsibilities of project managers?
KARAN ANAND: AI is definitely making us more productive. If I have to give a one-liner answer here, in my day-to-day administrative task, AI has been definitely helping our project managers a lot in making them more efficient. So, there are three pillars in which AI can impact project management. AI can do automation. AI can do assistance. And AI can do augmentation. So, it’s not what AI is doing for you, it’s what you can do with AI.
BILL YATES: Mm, yeah.
WENDY GROUNDS: Ah, very good.
KARAN ANAND: AI is here. It’s a module. It’s a software. And it’s a piece of code. That’s not going to do work for you. What you can do with AI, that’s important.
BILL YATES: Yeah.
KARAN ANAND: That’s how I call it.
The Impact of AI on Teams
BILL YATES: Talk to us about the practical aspects of AI usage and the impact or the shift that it’s had on how you work with your team.
KARAN ANAND: I can talk about a few common, very low-hanging fruits which I have been using. You know, my day-to-day life, I polish my emails before sending using AI. So, we have a plugin we call Gemini for Workspace. It’s I guess $15 per month. And what it does is, when you write an email, it’s in your draft. You can click a button, you want to polish it, you want to be formal, you want to make it casual. Which tone do you want to make it? I’m an immigrant here. English is not my first language. I make mistakes sometimes. AI has been helping. AI has been solving the language barrier.
But that’s some one of the major impact I’m seeing in my day-to-day life which I use almost every hour, making my emails better. Another thing I can think of of AI is project manager usually gets taken as a note-taker. Now all these meetings like a Zoom, Meet, all these video conferencing tools has a plugin to take a note summarizer. We don’t need a project manager to take our notes. AI can do those things.
There is going to be some strategic shift in the job duties of what project manager, program manager is doing. From now to one year or two year, I don’t know when the curve will get cut up. Plus, the curve will be different for different types of organization. For some organization they may have a curve sooner, like six months from now, someone has two years. It all depends on the maturity of an organization. And a sector, healthcare, financial, technology, energy sector, that curve is going to be separate.
Practical Training on AI
BILL YATES: Give us some practical advice to those project managers who are thinking, “Okay, I probably need some kind of training to raise my awareness of how I can use AI in some of these areas.” You know, you mentioned Gemini and the value that it’s bringing to you throughout the day. What’s some practical training on AI that some project managers should look at?
KARAN ANAND: I guess everyone should know not just what AI can do, how you can do with AI. That’s, again, I’m repeating the same thing because the practical content is very missing. If you don’t have a hands-on knowledge on what you can do with AI, then things are going to be very difficult. Second, I would recommend everyone to know about prompt entering.
Prompt entering is the basics on how you use AI. I can explain in a one-liner what prompt entering is. Let’s say you have any tool like an AI LLM tool, ChatGPT, or Gemini or anything. What we enter into the tool is a prompt. So prompt is going to do what you ask it. And it’s a duty of a consumer or an end user to give the right context to the prompt. If the context is missing, then it’s not going to give you the right result.
BILL YATES: Right.
KARAN ANAND: So, you need to give the right prompts. I recommend everyone to learn about these prompts, how to make your own prompts, how you can play with your prompts. Have a basic exercise like, for example, you have a big book. Just ask a Gemini or any LLM tool, “Hey, can you summarize this book for me in five sentences?” And see what the output gives. Then tell LLM, or I call LLM means any AI tool because I don’t want to advertise anyone here. So basically, you can always give, “Hey, Gemini, or hey, LLM tool, why did you give me this output, not this output?” Start playing with it. Once you start getting a hands-on knowledge, you will learn more than anything else.
BILL YATES: Yeah, yeah.
KARAN ANAND: Because the practical knowledge takes you a long way. And especially with these tools, you can ask them anything, and it’s very easy to give them context.
Stakeholder Prompts for AI
BILL YATES: Yeah. This is practical. This is great advice. And it’s so funny. I’m thinking of the conversation we just had recently on a previous podcast with Oliver Yarbrough, and we were talking about AI. And he said, “You really, you know, for project managers, think of AI as another stakeholder. It could be a member of your team, a very talented member of your team that never gets tired. You’ve got to get to know your stakeholder; right? You’ve got to know.”
KARAN ANAND: Right.
BILL YATES: So that’s such good advice that you have of you have to learn how to ask it questions, how to give it better prompts, how to give it the background that it needs to understand the context of what kind of response you’re trying to get from it. So, it kind of goes back in my head to we have to know our stakeholders in our projects. Now we have a stakeholder, we have a tool, powerful tool in AI. We have to know what’s the best way to use that tool. What are the best questions and prompts to ask?
KARAN ANAND: You need to tell the LLM the behavior of each stakeholder you have in your organization.
BILL YATES: Yeah.
KARAN ANAND: The LLM will not know how my director is, how my vice president is, how my CTO or CEO is. You need to tell CTO wants this, and my director wants this. How should I have a project plan for this?
BILL YATES: That’s right.
KARAN ANAND: Then AI is going to give me a project plan that’s going to make my vice president happy, my chief technical officer happy, and my director happy. You need to give the context because a lot of people struggle and suffer using AI, and they give up. They’re saying it’s hallucinating, it’s not giving consistent output because they’re not giving the right context.
BILL YATES: Yeah, that’s so true.
WENDY GROUNDS: Yeah. I have found that when I put something in, I have to be very specific. And sometimes it’ll give me something, and I’ll say, “No, change that word.” And, you know, you just have to keep getting better at the prompts, yeah.
BILL YATES: Mm-hmm, it’s true.
How AI can Identify Risks
WENDY GROUNDS: Let’s talk a little bit about risk management. How can you use AI tools to identify potential risks in your project life cycle? And how is that going to compare to traditional risk management?
KARAN ANAND: Again, traditional risk management was very human-oriented work, very manual work. There are a couple of ways in traditional project management. For example, someone may be just doing in his head, we did three projects, we had this risk, what the risk can be in the next project. Or if someone extraordinary project manager is keeping a track or a log of all the previous projects in some spreadsheet or some tool and having a column, what risk I got in this previous project. And based on those, we can have a categorization of projects, let’s say finance type of projects, we get these types of risks and then map it to the next finance project. That’s the traditional project management.
But with AI, this is going to be a more accurate, more precise, less time-consuming. We need to give the right historical dump of data again to AI, to LLM to set the context because they are not going to learn automatically what the risks of your organization are. LLM doesn’t know my people. LLM doesn’t know the risk of my organization that there could be iteration, there could be a resources crunch, there could be a cost high or anything like that. So, first of all, give the right context, the historical dump of data to LLM on these are the risks I got in my previous projects.
I’m going to derive and design a new project which will have this project plan. This is my project plan for the next project. What do you think the risk can come in my current project? AI is going to give you very detailed risk across the different factor, different dimensions, and sometimes human missed dimensions. It’s very easy for a normal human being to miss few dimensions because when we think, we only think from one or a couple of perspectives, and our thinking gets short‑sighted.
But when a machine thinks, they think far-sighted. They have a lot of context. In fact, they can search, they have access to all the information on the Internet and all the training data from which they can give you very good insight which a human being can’t give. All in all, what I’m trying to say, risk management using AI is very precise, accurate, more efficient, and going to save a lot of resources on a human-based project manager.
Accurate Historical Data
BILL YATES: With using AI for risk management, one of the keys is having accurate historical data that we can feed into the model. What advice do you have for project managers and their teams? How do I make sure I’ve got accurate historical data, and how do I feed that into the model?
KARAN ANAND: I guess that’s a very good question and interesting question. There are a couple of ways, it depends again on the organization structure, on how you are managing your projects. I’m not advertising any tool. Let’s say most organization I see use Atlassian Jira, at least in few organizations. If you have a Jira ticket or any ticketing mechanism to have a ticket for your workforce, you can use that to get a dump of data.
But again, it’s a duty of a project manager, a good project manager to make sure there is a Jira hygiene. I call it “ticket hygiene” because I see in a lot of places, tickets are assigned to one person, but person B is working, while ticket is assigned to person A. That’s a common workspace thing in any organization that the hygiene is not there. And the number of hours a person worked on a ticket are logged in properly if you are going in that direction.
Another dimension is have a clear risk, have a clear status of a ticket. When is it moving from in progress to fixed or done from not started? Because then we can get to the velocity of ticket. How much time does it take a ticket to get closed? That is a very good factor for calculating the risk because AI or LLM is going to see these types of tickets usually take more than expected time to close it out. Maybe I should call them as a risk for your next project because that’s going to hit on my critical path for the next project, as well. Again, hygiene of the data is very important. No matter if you’re having a manual dump of data, I don’t usually recommend a manual data because the chances of error are very high.
BILL YATES: Yeah.
KARAN ANAND: It’s always recommended to use any tool, whichever tool your organization, PMO, or other parts of your organization say, but make sure the data is very hygiene. And most of the tools have a way of exporting the data as a CSV.
BILL YATES: Right.
KARAN ANAND: You can always go to the dashboard or settings and export the data as a CSV or Excel. Just get a dump in your laptop and upload it to the LLM, and they’re going to do work for you.
BILL YATES: Yeah.
KARAN ANAND: And make sure it has all the columns. How much time did it take to close a ticket? What type of ticket is it? Who got it? All that information. The more you give a context to AI, the better your answer is going to be.
BILL YATES: Yeah, it’s so true. It gets back to the data at the lowest level. It needs to be, like you say, hygiene. And it needs to be accurate. It needs to be thorough. And then when you’re putting good data into the model, you can expect better results.
Data Security
All right. Big question here. What about security? When I’m uploading this data, what are considerations that I need to make within my organization regarding the security of that data?
KARAN ANAND: Again, beautiful question because a lot of people are not using cloud-based LLMs, I call it, for security or considerations. We saw some news about ChatGPT data hack, something long back. So yes, that’s why one of the reasons all these big players – Microsoft, Meta, Google – are coming with their LLMs because they have been considered reliable and secure and have already built a lot of trust with end users. That’s why it’s easy for them to sell their products to the end users.
Second of all, I will, I’m not saying you need to use these cloud-based services. You can deploy your in-house LLM, as well. So, there’s a website called Hugging Face. You can download an open source LLM and fine-tune it, and you can deploy your in-house LLM, in-house AI tool where you don’t have to upload a data to outside party. It depends on the type of organization. I’m seeing this pattern in financial and healthcare where they don’t prefer to upload their data outside to a third party, and they prefer in-house LLM.
I guess it’s the same concept as a cloud, or you are having your in-house servers. Some companies have their own data center infrastructure room in their company, run their own infrastructure; while some companies say, why do I have to carry such a big stack of hardware and have end of life and have a person to manage them? It’s going to be a big cost for me. But for some companies, the security and reliability is more important. They wanted to keep it in-house. It all depends on what your principles are, where you see your organizations. In my opinion, hybrid is going to be a future.
BILL YATES: Okay.
KARAN ANAND: Hybrid is a future where we will have some companies using one, some companies using other, but some companies using both.
Personal Project Experience using AI
WENDY GROUNDS: Do you have any personal examples? Have you used AI just to prevent potential risks or project delays? Have you had an experience like that where you’ve personally used it?
KARAN ANAND: So, like I said, I’m in the reliability organization, and our core job is to make sure production operations are up. Which means Google.com or Gmail.com are working fine for everyone. But we do projects. So in my project, if we get, let’s say interruption, some website is going down or having an outage, we leave the project, the engineers leave the project and work on the issue to keep the production up. Then when they leave the projects, the timeline gets impacted.
And it’s really hard to predict the timeline because you never know when the production will have an issue. Do anyone know when the Google.com will go down next time? No one knows. So, how my projects get impacted very usually depends on the alerts we get. The capacity is high. The latency is high. We need to go there, redirect our resources. All in all, this is challenges what I face, and how AI can help me is here.
I gave a historical dump of data again to AI. Hey, these are the 10 projects I solved in the last two years. And this much is projected time; this much time I actually took. One column for a projected time to complete; one column for actual time to complete. And then based on that, AI is helping me in the project planning for my 11th project. They can tell you; you should actually ask for these many resources in the next project.
BILL YATES: Yeah.
KARAN ANAND: If you’re going to take this much time. So, it’s helping me in a project planning and making sure I’m not slipping out of my timeline for my 11th project, basically giving me that right amount of resources I need for the next project. If AI wouldn’t help me, I would have slipped my 11th project, as well.
BILL YATES: Yeah.
KARAN ANAND: Yeah, it’s a very great opportunity where AI is helping us. But again, I need to give the right context to AI. I need to tell you are a project manager, you have been solving 10 projects, and your 10 projects were delayed, and this is the data. Can you tell me how much time will you take for the 11th project where I estimate this much time you will take? No, you are wrong.
AI in Resource Management
BILL YATES: Yeah. It’s good. Just in reading up on you ahead of time, you know, some of the things that really stood out to me was your use of AI to do resource management.
KARAN ANAND: Yeah.
BILL YATES: You know, I can envision like a chart or a spreadsheet showing the resources that are going to be working on your projects and seeing their skill set. And then, to your point, tracking their performance in the past and how much time your project loses if they get pulled into a different project or some other…
KARAN ANAND: Correct.
BILL YATES: …chaotic thing. Then you’ve got such a great rich set of data that AI can look at and help you plan out. You know, because for me, I tend to look at things as I put on really optimistic rose-colored glasses and think, oh no, I’ll have 100% of capacity for these team members. And then the reality hits. It always hits; right? So this is – it’s so helpful to have AI help you analyze and predict based on past experiences with resources.
KARAN ANAND: Correct. Again, then it boils down back to the same question. The cleaner the data you have, the more the AI can do for you.
BILL YATES: Yup.
KARAN ANAND: We manage the skill set of each employee where their skill expertise is. We have wanted to make their expertise, too. There are a few employees who say, I’m good in skill A, but I’m good in skill B, as well. So, I wanted to make sure they get an opportunity to work in skill B, as well. So, it all depends. We join multiple spreadsheets together to get an output of a data. Let’s say if we pull resources from a skill A, then who is another person whose skill A can fill it here. AI has been really helping us in data mining the resource allocation, as well. These are a few great examples where AI is helping us. That’s why I said AI is making us more efficient.
BILL YATES: Yeah.
KARAN ANAND: Apart from that, AI is opening a lot of opportunities for project managers, as well. AI is having a lot of problem about the resource efficiencies. How can we make AI more reliable? How can we make AI more privacy? All these things need project managers. So, it’s opening a can of opportunities, as well. It’s not just taking the jobs away.
Ren Love’s Projects of the Past
REN LOVE: Ren Love here with a glimpse into Projects of the Past; where we take a look at historical projects through the modern lens.
Today’s featured project is not entirely in the past – as it is a smaller component of the larger, ongoing Mars Exploration Program. We’ll be looking at the Mars Exploration Rover Mission, in which two robotic rovers were designed, built, & sent to explore the surface of Mars.
Design & construction of the two rovers was completed in 2003 – and the rovers were given the names ‘Spirit’ and ‘Opportunity’ as part of a student essay contest. The two rovers were launched in 2003 and arrived on the surface of Mars in January of 2024. The primary scientific objective of the mission was to investigate any clues that might lead to understanding any past water activity on Mars, in addition to documenting various minerals, soils, and rocks the rovers found along the way.
Initially, the mission was only scheduled to take 90 Martian Solar Days – which translates to just over 92 Earth Days – but both Spirit and Opportunity were active much longer than originally planned. Mars Rover Spirit stopped sending signals in March of 2010, while Opportunity was active until June of 2018. All in all, that 90 Sol Day mission turned into 16 years of exploration on the surface of Mars – an impressive feat of engineering and a victory for scientific discovery.
It’s impressive that the rovers survived so long past their expected expiration, but it was also costly. Original estimates for the cost of the mission came in at just over $800 Million USD – but there were five total mission extensions as the rovers remained operable. Each of those extensions added their own millions of dollars to the cost – and all told, the estimated final cost of the Mars Exploration Rover Mission was over a billion US Dollars.
So, was this project a success? Absolutely. Both rovers gathered and transmitted vast amounts of data – including meeting the mission goal of providing evidence that there was, in fact, past water activity on Mars. The last remaining rover, ‘Opportunity’ became part of the cultural zeitgeist, the rover was given the nickname ‘Oppy’ – and its last message to earth (which roughly translated to: My battery is low and it’s getting dark) inspired many fans to send back final messages of their own, thanking Oppy for its years of service.
Thanks for joining me for a look into Projects of the Past – I’m Ren Love. See ya next time.
AI and the Human Element
WENDY GROUNDS: AI is really good at automating our tasks, and it lacks leadership capabilities. Now, how do you strike that balance between you leveraging AI, you are using it for your projects, but it’s not capable of maintaining any human oversight, particularly when something is unpredictable, or if the project is changing rapidly.
BILL YATES: Like humans.
WENDY GROUNDS: Like humans, yeah. There’s that human element that AI can’t foresee.
KARAN ANAND: It’s missing, yeah.
WENDY GROUNDS: Yeah. How do you navigate that?
KARAN ANAND: Yes. I guess don’t over rely on AI. That’s where I would come from. You need to know what AI can do for you. I mean, it’s not going to automate in some things. It’s going to just assist or augment you. That’s why I divide into three pillars. It can do the basic groundwork, but human needs to sign off on some things.
You can’t say, AI, can you do this thing for me? Where high stakes are involved, human needs to get involved in such decision-making. We can’t just give to AI and go home. That’s not going to go happen. Human needs to involve. AI can just be like your secretary who can do some things for you, but can’t do everything. And then the final sign is from the human being only. A human being needs to review what the task AI has done, whether it has been done perfectly, and then sign it off. That’s how I see it.
BILL YATES: That’s the truth.
KARAN ANAND: Because, I mean, for example, when the stakes are high, especially in the decision-making, we can’t rely on AI. Similarly, if someone has to perform a surgery, we can’t rely on AI because human lives are important. Human lives matter. The stakes are high. We can’t have an AI there. That’s why the AI or the bots are still not there. We still need doctors and surgeons to perform a surgery.
BILL YATES: That’s so true.
KARAN ANAND: Otherwise, it’s not a difficult task for someone to make robots to do surgeries. Why? Because the stakes are high. But if the context is different in one, let’s say you want to do an eye surgery. What if the context of one eye is different, very unique from other eye. We can’t leave on robots. There may be, from five years from now, we can have robots who may do cleaning and some stuff for doctors. Or, hey, can you give me this, this scissor or any instruments? There may be a robot from few years now, but a doctor will be there, sign off.
BILL YATES: Mm-hmm.
KARAN ANAND: We need, when the stakes are high, we need human beings. We can’t just leave AI to do its stuff.
BILL YATES: That’s so true. And it’s interesting. I think about an AI model. Maybe our team’s been using it and feeding current project data into it and asking questions and, you know, helping with risk management. And then let’s say a change request comes along that’s pretty significant scope change request. So, you know, I might think as the project leader, hey, I’m going to feed this into the model. It’s got all my information about my project. I want to see what it says.
And it comes back with a recommendation, you know, reject it or accept it. Okay, that’s one opinion. I need to look at that holistically and see, okay, yeah, what else do I know about the key client or the key stakeholder? And what else is in this scope change request that maybe AI would not be aware of? You know, this is just another opinion. I’ve got to process that through my human brain or that of my team. Yeah.
KARAN ANAND: Right. You’re absolutely correct. A lot of time these answers are like, AI gives an answer like, “It depends,” like a usual consultant answer.
BILL YATES: Exactly.
KARAN ANAND: Depends on the context.
BILL YATES: Yup. Exactly.
WENDY GROUNDS: Don’t be vague. Ask AI.
KARAN ANAND: Yeah, they’re pretty vague sometimes. It depends on the context. That’s why you need human brain; right? You can’t leave things always on LLMs.
BILL YATES: Right.
Addressing Ethical Challenges
WENDY GROUNDS: So, let’s talk a little bit about ethics. What are some of the ethical challenges that AI brings to your industry, and how are you addressing ethical challenges?
KARAN ANAND: It’s getting better in terms of ethicals. In my company, we are huge on ethics. We make sure it gives the right output. In the past, a lot of other companies have seen biases get involved. Like AI has been biased about a decision, hallucinations. It doesn’t give the same consistent output across the different times. But things are getting better. We have seen those things in the past. I can share a couple of examples here. One example could be like building on the similar example, how many resources do I need for my 11th project, like I gave in the past. If you ask one time, it tells you, you may need this time. Then if you prompt again after 30 minutes with the same context, same prompt, the output could differ. That’s where the problem sometimes has been arise in the past.
But things are getting better now. There’s a lot of things getting improved in the space, as this is pretty new space, last one or two years. This has been exposed to general public a lot. And when you expose these products, you get to know the weaknesses. As long as the companies are investing in this and making things better, these things will be very solid one day because we have seen a lot of things like different biases, like gender bias, race bias, all these things in the past from these LLM tools, which are not right.
And companies have acknowledged that, and they’re getting better. A lot of times we haven’t filtered and blocked words in these tools while giving an output like these words are blocked. Or if you tell LLM, hey, should I buy the stock of company A, then they say, I’m not a financial advisor.
A Mid-Mortem Assessment
BILL YATES: Right. Okay, we’re going to shift just a little bit. We’ve been focused so much on AI, Karan. And again, I saw some of the writing that you’ve done, some of the blog posts and other things that you’ve written. One of the topics really jumped out at me, and I thought, I want to ask him to explain this further with our listeners because it’s so practical. And it has to do with what you refer to as a mid-mortem assessment. So, there’s, you know, before we launch a project, sometimes we’ll do a pre-mortem.
And then most project managers are familiar with doing a post-mortem or a review session at the end of a project. I love your idea of a mid-mortem, which is, okay, why do we wait until the very end? Let’s have a check-in. Let’s dedicate time to this, have a feedback session. And you refer to it as a mid-mortem. Share more about that. What’s the big idea with it, and what are the benefits?
KARAN ANAND: I guess it’s a very important, mid-mortem is an important aspect for large programs which are running for many years, more than a year, because when a large project gets run, it’s easy to get lost,
BILL YATES: Yes.
KARAN ANAND: Easy to lose track. And sometimes the project managers working there may not be as efficient as the stakeholders want, or the stakeholders are not efficient, some of the stakeholders. It’s good to set the right alignment because, as each day is not a rainy day, similarly each day is not a sunny day.
BILL YATES: Right.
KARAN ANAND: And there are always ups and downs in the projects, as well. So, it’s always important to know how you can get the things on track. If you’re at a down-level stage, how can you, first of all, assess whether you are up or down?
BILL YATES: Yup.
KARAN ANAND: If you are down, how can it bring things back up? I started for my project, which was three-years-long project, and I was in the middle of a project one year. I wanted to assess how we are doing, whether we are doing correct. Do we need to change anything on what we are doing? If yes, what? If not, why not?
BILL YATES: Right.
KARAN ANAND: To have a sturdy retrospective, go back, go front, whether I’m on track or not. So that’s where the thesis, I built this mid-mortem, and it really helped. I put everything on what we did till now. What’s our plan till now? What things went well? What things didn’t went well? In this last one year, how we plan ahead in the next two years, what are our plans to achieve the things, how we plan to achieve, then to come up with some recommendations on how you can fix up some things which you think the risks are. Again, coming up with risk for the next two years. What are the threats to your projects, and how you can fix those threats before they actually hit? Because why?
Because when you do a pre-mortem, the ambiguity, at least in large-scale complex programs, ambiguity is so high. You don’t know the actual stakeholders. Stakeholders may change across the three years. Their opinions get changed over the three years as you get more data. So, the pre-mortem is not sometimes that helpful as mid-mortem because we have built a context. We have been in the game for a year.
BILL YATES: Yes.
KARAN ANAND: You know the things, how things are going, how things will go. So mid-mortem has been more helpful to me than pre-mortem.
BILL YATES: Yeah, that is so rich. You bring out so many good points. So, the longer my program or project, the greater the need for me to have this mid-mortem. Any change in stakeholders, especially key stakeholders who define the scope, the requirements that we’re hitting, if they’ve moved on or changed or whatever, or at least even if they’re intact, you know, why not ask them again? Can we confirm our benefits? Should we still have the same goals in mind? Yeah, it’s such a good thing.
And I see kudos to you for recognizing the human side of teams, too. I mean, as a leader I could start to lose alignment, and I’m sure my team members were, too. This gives us a chance to refocus on what are the goals of our project. Have they changed? No. Are we still doing this the proper way? Yes. Great. We’re back on it. Let’s do it.
KARAN ANAND: Again, to call out the ChatGPT, no one knew ChatGPT two years back. Now everyone knows AI ChatGPT. Things have been changing so fast in the landscape. We have to keep updated and keep our projects updated with the recent trends and advancements to see if we can leverage anything. Are they impacting my scope? It’s a very bad idea to keep things intact, frozen for three years, and working in a black hole for three years.
BILL YATES: I agree. Yeah, it’s not – it’s not practical. You’re right. Too many things change.
KARAN ANAND: Not a practical idea, yeah.
BILL YATES: Yeah, yeah.
KARAN ANAND: Because the world is moving very fast from the last few years. It’s been very fast.
BILL YATES: Mm-hmm.
KARAN ANAND: Yeah.
Get in Touch with Karan
WENDY GROUNDS: We’ve really appreciated talking with you, Karan. Can our listeners get in touch with you, or can they find out more about what you do? Where should they go?
KARAN ANAND: LinkedIn is a pretty decent spot. Have a healthy balance. Feel free to connect me on the LinkedIn for any questions. I’m happy to help or give an advice on any topic. Would love to.
BILL YATES: Well, we so appreciate the chance to connect with you and just hear more. I mean, you are right there in the trenches. You’re using AI. You’re impacting resource allocation across dozens and dozens of resources. You guys are using this to look into risk management and to really go next-level with AI. And I appreciate your practical examples, everything from Gemini to some of the uses in risk management. And it’s an exciting time, and I appreciate your attitude as a project leader. You know, we don’t need to be afraid of this. We just need to see it as a tool, think about the limitations with AI, and embrace it and dive into it. Thank you for your words on that.
KARAN ANAND: Thank you for having me here. Yeah, again, it was a great conversation. You asked very nice questions hitting on the spot about the plus and minus of AI. And definitely I agree with you. We should embrace the AI, the change. Change is always hard for any human being. If anyone is waking up at 6:00 a.m., you ask them to wake up at 4:00 a.m., change is hard, similarly.
WENDY GROUNDS: Yeah.
BILL YATES: Yes. Yeah, great example.
KARAN ANAND: Similarly, change of AI is hard, too. Right? So, but we need to embrace the change, learn with this change, be vulnerable because in the end we are all going to grow with this change.
BILL YATES: That’s right.
KARAN ANAND: If we don’t embrace this change, then change will eat us.
BILL YATES: That’s true. Thank you so much.
WENDY GROUNDS: Thank you.
Closing
WENDY GROUNDS: That’s it for us here on Manage This. Thank you for joining us. You can visit us at Velociteach.com, where you can subscribe to this podcast and see a complete transcript of the show.
Now for our chance to give back. You’ve earned your free PDUs by listening to this podcast. To claim them, go to Velociteach.com. Choose Manage This Podcast from the top of the page. Click the button that says Claim PDUs and click through the steps.
Until next time, stay curious, stay inspired, and keep tuning in to Manage This.
Leave a Reply