Beyond the Chatbot: AI and ML Investment Trends w/ Asad Malik

In this episode of Financial Perspectives, host Tanya Tang sits down with Asad Malik, CFA, Head of ML Finance & Infrastructure Strategy at Google, to discuss how artificial intelligence and machine learning are transforming finance, investing, and productivity.

Asad breaks down the surge in AI investment since 2022, the rapid growth in infrastructure spending, and why scaling laws and vertical integration are redefining the economics of innovation. He explains why AI may be overhyped in the short term but underappreciated in the long run and highlights key areas of opportunity—from drug discovery and automation to AI-driven productivity efficiencies.

Listeners will also hear practical advice for finance professionals on adopting AI tools, maintaining essential analytical skills, and leveraging AI as a powerful productivity advantage.

Tune in for a strategic financial perspective on AI’s impact on finance, investment strategy, and the future of work.

Episode Transcript

Lindsey Helman: Hello and welcome to Financial Perspectives, a CFA Society San (00:00):
undefined

Lindsey Helman: Francisco podcast, where we interview and discuss trends (00:05):
undefined

Lindsey Helman: with leaders from across the investment and finance industry. (00:09):
undefined

Lindsey Helman: This month, our host Tanya (00:13):
undefined

Lindsey Helman: Suba-Tang, membership director (00:14):
undefined

Lindsey Helman: with CFA Society San Francisco, (00:17):
undefined

Lindsey Helman: had the pleasure of speaking (00:19):
undefined

Lindsey Helman: with Asad Malik, CFA, head of (00:20):
undefined

Lindsey Helman: Machine learning Finance and (00:24):
undefined

Lindsey Helman: infrastructure strategy at (00:25):
undefined

Lindsey Helman: Google. (00:27):
undefined

Lindsey Helman: Listen in as they discuss what investors should know about the (00:27):
undefined

Lindsey Helman: AI boom and the future of machine learning finance. (00:31):
undefined

Tanya Tang: Good morning Asad, how are you? (00:33):
undefined

Asad Malik: Good morning Tanya. I'm good. (00:36):
undefined

Asad Malik: How are you? (00:37):
undefined

Tanya Tang: I'm great. (00:38):
undefined

Tanya Tang: Thank you so much for joining me at today's podcast episode. (00:39):
undefined

Tanya Tang: I'm very excited to have our conversation started. (00:43):
undefined

Tanya Tang: And because we're talking about something that I think it's more (00:47):
undefined

Tanya Tang: than a buzzword now. (00:51):
undefined

Tanya Tang: ML and AI, right? (00:52):
undefined

Tanya Tang: It's been going on for quite some time, but we're going to (00:54):
undefined

Tanya Tang: take a different approach and perspective about it. (00:57):
undefined

Tanya Tang: And you know, with your background as leading ML finance (00:59):
undefined

Tanya Tang: strategy for Google, who better to talk to than you about it? (01:03):
undefined

Asad Malik: Tanya, first of all, thank. (01:07):
undefined

Asad Malik: You for having me on your Financial Perspectives podcast. (01:08):
undefined

Asad Malik: And I'm I hope I can share some personal views that will help (01:11):
undefined

Asad Malik: our listeners think about all the recent ML funds. (01:14):
undefined

Asad Malik: So it's super exciting to be here. (01:16):
undefined

Tanya Tang: Great. (01:19):
undefined

Tanya Tang: So let's start a little bit of a background. (01:20):
undefined

Tanya Tang: You know, when we were talking before the podcast, you had (01:23):
undefined

Tanya Tang: mentioned significant changes in AI and ML investment trends (01:26):
undefined

Tanya Tang: since twenty twenty two. (01:31):
undefined

Tanya Tang: Wow. (01:32):
undefined

Tanya Tang: So long ago. (01:32):
undefined

Tanya Tang: could you share a bit more about (01:34):
undefined

Tanya Tang: what you think is driving these (01:35):
undefined

Tanya Tang: changes? (01:36):
undefined

Asad Malik: Yeah. (01:37):
undefined

Asad Malik: For sure. (01:38):
undefined

Asad Malik: It feels like yesterday, to be honest. (01:39):
undefined

Asad Malik: It's moved so fast. (01:41):
undefined

Asad Malik: I mean, AI research, it's been advancing for decades, but I (01:43):
undefined

Asad Malik: think the big bang moment was really in late twenty twenty (01:47):
undefined

Asad Malik: two, when millions of people got their hands on some of the most (01:50):
undefined

Asad Malik: powerful large language models. (01:54):
undefined

Asad Malik: Right? (01:56):
undefined

Asad Malik: And maybe it was a pandemic that locked us down and we just (01:56):
undefined

Asad Malik: needed someone to talk to. (02:00):
undefined

Asad Malik: But with AI chatbots, you could (02:01):
undefined

Asad Malik: ask questions the way you would (02:03):
undefined

Asad Malik: with a human on the other end, (02:05):
undefined

Asad Malik: and you'd get these detailed (02:06):
undefined

Asad Malik: responses on any range of (02:08):
undefined

Asad Malik: topics. (02:09):
undefined

Asad Malik: even the flow of information, (02:11):
undefined

Asad Malik: the way you, had this back and (02:12):
undefined

Asad Malik: forth dialogue. (02:14):
undefined

Asad Malik: It was the same way as if you were interacting with an expert. (02:14):
undefined

Asad Malik: And then you have image (02:18):
undefined

Asad Malik: generation and now video (02:19):
undefined

Asad Malik: generation, which has only (02:21):
undefined

Asad Malik: further captured our (02:22):
undefined

Asad Malik: imagination. (02:23):
undefined

Asad Malik: Like if you wanted a visual or a clip of something, you could (02:24):
undefined

Asad Malik: just create it in an incredibly short amount of time. (02:27):
undefined

Asad Malik: And so that massive public (02:30):
undefined

Asad Malik: interest, it's really sparked (02:32):
undefined

Asad Malik: this surge of investments in (02:34):
undefined

Asad Malik: this area. (02:35):
undefined

Asad Malik: And it's only grown since then. (02:36):
undefined

Asad Malik: So the early chatbots, they were (02:38):
undefined

Asad Malik: very cool, but you could still (02:39):
undefined

Asad Malik: see flaws in the quality of the (02:42):
undefined

Asad Malik: responses. (02:43):
undefined

Asad Malik: Right. (02:44):
undefined

Asad Malik: And in some cases you were getting answers that were just (02:44):
undefined

Asad Malik: flat out wrong. (02:46):
undefined

Asad Malik: But there was and there still is a strong belief in scaling laws (02:47):
undefined

Asad Malik: that if you made the models larger, more complex, and you (02:52):
undefined

Asad Malik: invested more in the computational power behind them, (02:56):
undefined

Asad Malik: you could create an even better model, something that was more (02:59):
undefined

Asad Malik: accurate and helpful. (03:02):
undefined

Asad Malik: And so you take that belief in (03:04):
undefined

Asad Malik: scaling laws with some notion of (03:06):
undefined

Asad Malik: monetization at the end of the (03:07):
undefined

Asad Malik: rainbow, and you get billions of (03:09):
undefined

Asad Malik: dollars now being poured into (03:11):
undefined

Asad Malik: this industry. (03:12):
undefined

Tanya Tang: So I'd love to get your thoughts (03:13):
undefined

Tanya Tang: on all the funding going on into (03:14):
undefined

Tanya Tang: this industry. (03:16):
undefined

Tanya Tang: And, you know, what are the (03:17):
undefined

Tanya Tang: areas investors should keep an (03:19):
undefined

Tanya Tang: eye on? (03:21):
undefined

Asad Malik: Yeah, that's. (03:22):
undefined

Asad Malik: Actually a really important question. (03:23):
undefined

Asad Malik: I mean, we talked about scaling laws. (03:24):
undefined

Asad Malik: you need to grow compute (03:26):
undefined

Asad Malik: infrastructure to improve large (03:28):
undefined

Asad Malik: language models and create more (03:29):
undefined

Asad Malik: helpful products. (03:30):
undefined

Asad Malik: So that's number one. (03:31):
undefined

Asad Malik: Number two, you also need (03:32):
undefined

Asad Malik: investments in infrastructure to (03:33):
undefined

Asad Malik: serve those products to the (03:35):
undefined

Asad Malik: consumers, right. (03:36):
undefined

Asad Malik: This is the capacity you need to (03:37):
undefined

Asad Malik: bring that model to billions of (03:39):
undefined

Asad Malik: users. (03:40):
undefined

Asad Malik: And then number three, you need the talent to implement all this (03:41):
undefined

Asad Malik: and drive progress. (03:45):
undefined

Asad Malik: So when you add all those three (03:46):
undefined

Asad Malik: up you get these really (03:48):
undefined

Asad Malik: expensive builds. (03:49):
undefined

Asad Malik: And I think the industry as a whole is likely spending around (03:51):
undefined

Asad Malik: four x this year than what it did back in twenty twenty three. (03:54):
undefined

Asad Malik: Building all this infrastructure also takes time. (03:57):
undefined

Asad Malik: So the investments in growth you (04:00):
undefined

Asad Malik: see today are still constrained (04:01):
undefined

Asad Malik: since the demand showed up (04:04):
undefined

Asad Malik: yesterday. (04:05):
undefined

Asad Malik: But you can tell from the large (04:05):
undefined

Asad Malik: project and some of the deal (04:07):
undefined

Asad Malik: announcements that this is only (04:08):
undefined

Asad Malik: going to continue growing in the (04:09):
undefined

Asad Malik: future. (04:11):
undefined

Asad Malik: Another area for investors to (04:11):
undefined

Asad Malik: observe, I think, is that the (04:13):
undefined

Asad Malik: leading companies in this field, (04:15):
undefined

Asad Malik: at different stages of the AI (04:17):
undefined

Asad Malik: economic chain, they're trying (04:18):
undefined

Asad Malik: to become more vertically (04:20):
undefined

Asad Malik: integrated. (04:20):
undefined

Asad Malik: So some are already there, but basically it means that you own (04:21):
undefined

Asad Malik: the infrastructure, the chip development, the research and (04:25):
undefined

Asad Malik: the end products. (04:30):
undefined

Asad Malik: That's a very important way of integration that helps you drive (04:31):
undefined

Asad Malik: efficiencies at scale. (04:35):
undefined

Asad Malik: So given all these billions of investments, the expectations (04:37):
undefined

Asad Malik: from the industry are certainly very, very high. (04:40):
undefined

Asad Malik: And so this is my view. (04:43):
undefined

Asad Malik: And I'm borrowing from leaders from the industry that are much (04:44):
undefined

Asad Malik: more knowledgeable about this than I am. (04:48):
undefined

Asad Malik: But I think AI is probably overhyped in the near term and (04:50):
undefined

Asad Malik: significantly underappreciated in the long term. (04:54):
undefined

Asad Malik: Right? (04:58):
undefined

Asad Malik: Like the model improvement, the product development, and more (04:58):
undefined

Asad Malik: importantly, just adoption within the industry and (05:01):
undefined

Asad Malik: consumers, it will take time. (05:03):
undefined

Asad Malik: But all that said, I think AI is (05:06):
undefined

Asad Malik: still one of the most (05:07):
undefined

Asad Malik: significant transformations of (05:08):
undefined

Asad Malik: our lifetime. (05:10):
undefined

Asad Malik: There's a lot to be excited about. (05:10):
undefined

Asad Malik: And so for investors, you really (05:13):
undefined

Asad Malik: have to strike that balance (05:14):
undefined

Asad Malik: between being bold but also (05:15):
undefined

Asad Malik: responsible. (05:18):
undefined

Tanya Tang: I appreciate all that feedback. (05:18):
undefined

Tanya Tang: I totally agree with you. (05:20):
undefined

Tanya Tang: And I could see exactly where you're coming from. (05:21):
undefined

Tanya Tang: you know, in your opinion, what (05:24):
undefined

Tanya Tang: are some of the most promising (05:25):
undefined

Tanya Tang: products currently being (05:27):
undefined

Tanya Tang: developed in the AI space that (05:28):
undefined

Tanya Tang: you could see Reshape (05:29):
undefined

Tanya Tang: industries? (05:31):
undefined

Asad Malik: Yeah, there's a lot of promising (05:32):
undefined

Asad Malik: products being developed at the (05:34):
undefined

Asad Malik: moment, and I think some of the (05:35):
undefined

Asad Malik: most exciting ones, in my (05:37):
undefined

Asad Malik: opinion, are really in number (05:38):
undefined

Asad Malik: one drug discovery, number two (05:41):
undefined

Asad Malik: in automation. (05:43):
undefined

Asad Malik: And number three, I would say general productivity. (05:45):
undefined

Asad Malik: So if you look at drug discovery, the process today (05:47):
undefined

Asad Malik: relies heavily on lab experiments like these can take (05:51):
undefined

Asad Malik: over a decade and about three billion dollars on average to (05:54):
undefined

Asad Malik: bring a new drug to the market. (05:58):
undefined

Asad Malik: AI could accelerate that by ten x, and that's very exciting. (05:59):
undefined

Asad Malik: And you can expand that to (06:04):
undefined

Asad Malik: general scientific research as (06:05):
undefined

Asad Malik: well, like the amount of PhD (06:06):
undefined

Asad Malik: hours it takes to make (06:08):
undefined

Asad Malik: breakthroughs could be (06:09):
undefined

Asad Malik: dramatically reduced. (06:10):
undefined

Asad Malik: And the outcomes for that for the industry and also humanity (06:12):
undefined

Asad Malik: as a whole are just incredible. (06:15):
undefined

Asad Malik: Number two, I would say automation. (06:18):
undefined

Asad Malik: So you have these predictive models and there are a range of (06:20):
undefined

Asad Malik: industries that will benefit from these models that improve (06:24):
undefined

Asad Malik: automation and drive significant change in processes. (06:26):
undefined

Asad Malik: So, for example, being able to forecast customer demand, being (06:29):
undefined

Asad Malik: able to predict when things will fail, improve quality control, (06:33):
undefined

Asad Malik: detect fraud and so much more. (06:37):
undefined

Asad Malik: I think this is going to hugely (06:39):
undefined

Asad Malik: revamp how a lot of industries (06:40):
undefined

Asad Malik: think about their operations (06:42):
undefined

Asad Malik: today. (06:43):
undefined

Asad Malik: And number three, I would say general productivity. (06:44):
undefined

Asad Malik: So writing code, creating (06:46):
undefined

Asad Malik: spreadsheets, putting together (06:49):
undefined

Asad Malik: presentations. (06:50):
undefined

Asad Malik: There's so much time that goes into that. (06:51):
undefined

Asad Malik: You have a rough idea of the output you're looking for, but (06:54):
undefined

Asad Malik: you need to go through steps of one to one hundred to get there. (06:56):
undefined

Asad Malik: So having AI that can help you skip most of those basic steps (06:59):
undefined

Asad Malik: involved is really going to unlock a lot of our time that we (07:03):
undefined

Asad Malik: can spend thinking more about the analysis or the end product. (07:07):
undefined

Asad Malik: And so there's a lot more that I haven't covered, but these are (07:11):
undefined

Asad Malik: some of the most important ones from my perspective. (07:13):
undefined

Tanya Tang: I think that's a really fantastic feedback. (07:15):
undefined

Tanya Tang: you know, these are great (07:18):
undefined

Tanya Tang: sectors to really look at and (07:19):
undefined

Tanya Tang: and see what how AI can kind of (07:21):
undefined

Tanya Tang: help, as you said, accelerate (07:23):
undefined

Tanya Tang: things. (07:24):
undefined

Tanya Tang: and, you know, AI has been (07:25):
undefined

Tanya Tang: around for quite some time, as (07:28):
undefined

Tanya Tang: we said in this podcast, but (07:29):
undefined

Tanya Tang: there's still some hesitancy (07:30):
undefined

Tanya Tang: from a lot of people in the (07:31):
undefined

Tanya Tang: public, whether in the industry (07:33):
undefined

Tanya Tang: or outside. (07:34):
undefined

Tanya Tang: What advice would you give, (07:36):
undefined

Tanya Tang: whether students or (07:37):
undefined

Tanya Tang: professionals who are listening (07:39):
undefined

Tanya Tang: in and are still hesitant to (07:41):
undefined

Tanya Tang: incorporate AI into their daily (07:43):
undefined

Tanya Tang: work routines? (07:45):
undefined

Tanya Tang: And really, how how do you get started? (07:45):
undefined

Asad Malik: Yeah, it's funny you say that (07:49):
undefined

Asad Malik: because I would put myself in (07:50):
undefined

Asad Malik: the hesitant professional (07:52):
undefined

Asad Malik: category. (07:53):
undefined

Asad Malik: Um, I'm still learning how to make the most of these new (07:54):
undefined

Asad Malik: available tools as well. (07:57):
undefined

Asad Malik: But I think it starts with intention and curiosity. (07:59):
undefined

Asad Malik: You have to ask yourself, is (08:02):
undefined

Asad Malik: there any part of what I'm doing (08:03):
undefined

Asad Malik: right now that could be done (08:05):
undefined

Asad Malik: with AI? (08:06):
undefined

Asad Malik: Maybe if you have a sibling or (08:08):
undefined

Asad Malik: your new manager with a team now (08:09):
undefined

Asad Malik: reporting to you, this will (08:11):
undefined

Asad Malik: sound familiar, but you almost (08:12):
undefined

Asad Malik: have to retrain yourself to (08:13):
undefined

Asad Malik: think from a perspective of hey, (08:15):
undefined

Asad Malik: what can I delegate and get done (08:16):
undefined

Asad Malik: for me? (08:18):
undefined

Asad Malik: So yes, I could pull this presentation together, but can I (08:19):
undefined

Asad Malik: get AI to do it for me? (08:23):
undefined

Asad Malik: I could read through all these reports, but can I get AI to (08:24):
undefined

Asad Malik: summarize and maybe help me focus on the most important (08:27):
undefined

Asad Malik: parts of it. (08:30):
undefined

Asad Malik: And I would say really try out all the available products, see (08:31):
undefined

Asad Malik: what works for you. (08:34):
undefined

Asad Malik: So that said, I still think it's (08:36):
undefined

Asad Malik: incredibly important for (08:38):
undefined

Asad Malik: students and professionals to (08:39):
undefined

Asad Malik: develop their underlying core (08:40):
undefined

Asad Malik: skills, right? (08:41):
undefined

Asad Malik: Like, if you don't know how to code or create a research report (08:42):
undefined

Asad Malik: and so on, it'll be hard for you to understand what kind of (08:44):
undefined

Asad Malik: outputs you're getting from AI and how to best filter those for (08:47):
undefined

Asad Malik: the final product. (08:50):
undefined

Asad Malik: I'd also recommend caution on AI outputs. (08:52):
undefined

Asad Malik: It's important to continue sense checking. (08:54):
undefined

Asad Malik: You want to think critically (08:56):
undefined

Asad Malik: about the answers that you're (08:58):
undefined

Asad Malik: getting. (08:59):
undefined

Asad Malik: But I think at the end of the day, really think of AI as a (09:00):
undefined

Asad Malik: powerful productivity boost and intentionally find areas to (09:02):
undefined

Asad Malik: leverage it and free up your time and bandwidth. (09:06):
undefined

Tanya Tang: Those are really some great feedback and thank you so much (09:08):
undefined

Tanya Tang: Asad for your time. (09:11):
undefined

Tanya Tang: I know it's been a very busy (09:12):
undefined

Tanya Tang: time for you personally, so I (09:14):
undefined

Tanya Tang: really appreciate all the (09:16):
undefined

Tanya Tang: insights and feedback you (09:17):
undefined

Tanya Tang: provided us, and I'm sure our (09:18):
undefined

Tanya Tang: listeners are going to have a (09:20):
undefined

Tanya Tang: lot to take away and digest in (09:21):
undefined

Tanya Tang: this episode. (09:23):
undefined

Asad Malik: Awesome. Thanks, Tanya. (09:24):
undefined

Lindsey Helman: Thank you to this month's guest, Asad Malik, for joining us today (09:25):
undefined

Lindsey Helman: and shedding light on the evolving world of AI and machine (09:29):
undefined

Lindsey Helman: learning in finance. (09:32):
undefined

Lindsey Helman: Join us next time for another (09:33):
undefined

Lindsey Helman: Financial Perspectives episode (09:35):
undefined

Lindsey Helman: airing on the last Tuesday of (09:37):
undefined

Lindsey Helman: the month. (09:39):
undefined