AI Q&A: Scott Penberthy, Google
Artificial intelligence (AI) is the buzz-phrase these days – especially in finance.
How can the buy side better utilize it in making smarter and more efficient trading decisions. For the sell side, the question is how can AI help them create better automated tools that can attract order flow and buy-side business.
Enter Google, which has been employing AI for years making life easier outside Wall Street – helping ordinary people find the best price on travel to arcane information.
Recently, Scott Penberthy, Director of Applied AI at Google, spoke with Traders Magazine’s editor John D’Antona Jr. about AI and the capital markets. Penberthy discussed the future trends in AI projects, the kind of AI project more financial institutions will take on and best practices for evaluating said projects.
TRADERS MAGAZINE: How do you see AI evolving?
Scott Penberthy: Think of AI like databases from the 1980s. Back then, databases were new, fresh and cool. But today, we don’t ask whether an application or system is “database-powered.” Databases are just a tool — a powerful and necessary one — in building software systems.
AI is moving that way, too. Systems powered by AI will handle more inbound data than humans, make better predictions than we could by ourselves, and often generate novel experiences for our customers. AI-powered software will get better with use, as it learns and adapts. This will map to business processes, which will continuously improve.
TM: What AI projects do you predict financial institutions will tackle next?
Penberthy: Financial institutions have leveraged machine learning (ML) and quantitative analysis for years, and we’ve already been working with financial institutions for quite some time to leverage AI for many different business cases. The global financial institution ING infused AI into its early warning system for credit risk analysts to vet potential risk exposure for clients. Canadian data company Flinks created a risk scoring engine with AI so that its customers could present best credit offers to users. And Grasshopper, a tech-based trading company out of Singapore, has leveraged our Cloud Machine Learning Engine to power large machine learning (ML) tasks.
What’s new is that computers can now see, read, hear, write and speak — improving exponentially with compute power and advances in algorithms. Computers can make predictions with incredible speed and accuracy.
We’re seeing core business problems reframed as predictions, especially in areas where a shortage of talent impacts customer experience. For example, contact center AI can handle inbound customer inquiries via voice, chat, email and imagery, augmenting with machines that predict customer intention. In another area, companies suffer from data being trapped in scanned documents, PDFs, and unstructured text. With document understanding AI, computers can read these documents and predict the data to extract.
TM: Who is driving these projects – the banks? Investors such as large pension funds and money managers?
Penberthy: It’s really the entire industry. Investing in AI is a hallmark of leadership in financial services, in addition to other industries.
TM: What are the best practices you recommend to financial firms when evaluating AI solutions?
Penberthy: Start with core business. How can you reframe your biggest challenges as a prediction problem? After that, figure out what data you’d need to make the prediction — where you think the “signal” sits in giving you sufficient insight to make the prediction. Use that insight to pull relevant data into the cloud, clean it and create dashboards. Then, begin iterating on AI that automates predictions. BigQuery, BigQuery ML and AutoML tables are a great place to start on Google Cloud Platform.
TM: Are we at a point where AI will replace employees?
Penberthy: AI can empower — not replace — employees by relieving them of some of the more mundane tasks and freeing them to focus on more impactful efforts. AI can automate some tasks and handle the data, without the need to increase a workforce. Additionally, AI can help employees become upskilled and as a result have more meaningful jobs and long-term careers. AI is a tool to be leveraged, rather than a new workforce.
TM: Are the benefits of AI being transferred to consumers or does it benefit financial institutions?
Penberthy: A growing number of financial institutions are applying AI to customer advice and interactions, laying the groundwork for self-driving finance. Customers will vote with their wallet, preferring businesses that adapt to their changing needs, faster, and more accurately than others.
TM: What does the future look like for Google Cloud and AI, particularly in the financial industry?
Penberthy: Our mission has always been to make the world’s information universally accessible and useful and now, we’re turning this focus to the world’s business information as well.
Financial services are replete with written documents and time series data. Reviewing documents is tedious and error prone, yet is a necessary task in regulation compliance, customer service, call centers and many business processes. We see these tasks as a perfect use case for AI, reducing tedium, improving accuracy, and improving the user experience at a lower cost.
We’d also like to make predictions and find anomalies in time series data. This mitigates financial crime, improves compliance, and reduces risk.
Data extraction and integration is the second stage of a digitization process.
Increased electronification has created useable and accessible real-time and historic trade data.
The fintech uses data so institutions can assess the environment impact of their portfolios.
Working with Riskfuel will reduce reliance on slow and expensive financial models.
The platform uses AI to match experts and users and digitize the research process.