
Conning ConnText is a quarterly podcast that features our firm’s view of capital markets, trends and investment strategies for the insurance industry, hosted by Rich Sega, Conning’s Global Chief Investment Strategist. Rich will also host a guest to provide additional insight about a particular asset class, investment idea or insurance industry trend.
Conning ConnText is intended to provide just that: context for insurance investing. Conning’s deep understanding of the investment environment and the insurance industry helps us to serve the unique needs of insurers.
Here is our latest Conning Conntext podcast:
October 2023: Applying AI to the Investment Process with Ole Jorgensen, Research Director, Global Evolution
In the Conning ConnText podcast for the fourth quarter, Rich Sega, Global Chief Investment Strategist, discusses the challenges to a resilient U.S. economy. He is joined by Ole Jorgensen of Global Evolution, who discusses artificial intelligence and how it can be applied to the investment process.
October 2023 Conning ConnText Podcast Transcript
OPENING
Rich Sega: I’m Rich Sega. Welcome to the Conning ConnText podcast for the fourth quarter of 2023.
Our aim in these podcasts is to cover recent developments and project possible outcomes for the economies and markets in which our clients are engaged. And we’ll do that in today’s podcast, which will feature a discussion about trends in artificial intelligence with Ole Jorgensen from Global Evolution, Conning’s emerging markets affiliate.
But we also know that these concerns seem quite secondary to the horror recently witnessed in Israel, and the ongoing conflict that promises even more anguish. These and other forces combine to shape the current state of global economies and markets; our discussion of those other forces should not be taken to diminish our concern for the massive human suffering which we now observe.
ECONOMIC/CAPITAL MARKETS OVERVIEW
RS: The U.S. economy has proven quite resilient, helped by the consumer. There are signs that the U.S. consumer might be pulling back a bit and that higher rates are beginning to affect at least the lower tier of income earners, and we believe that sometime in the first half of next year, the rolling recession will finally hit consumers more broadly. To date though, it has been a Goldilocks economy with everything just about right - not too hot or cold, in labor, manufacturing, and housing.
Now there are three major bearish threats facing investors as we move through Q4, each inflationary in the near term, and negative for growth over time. The first is of course the geopolitical risk that global markets must absorb. It began with the Russia invasion of Ukraine last year and continues with the attack on Israel. The emergence of a two-front war in Israel poses threats not only to Israel’s domestic and tourism-based economic sectors, but to the broader region and to energy markets. Energy markets are and have been the first to react, but the threat of a broader war and the protracted strain on trade will impose costs on global GDP and put pressure on consumer prices.
A second factor is persistent inflation itself, causing central banks to push back until “something breaks,” which of course will be the economic expansion. Fallout from that would suggest an end to the decade-long dominance of U.S. markets, a weaker U.S. dollar, and an opportunity for non-U.S. investors, emerging market issuers - but probably not Europe nor the UK. which are still struggling. It’s enough to take the foam off their Oktoberfest beer.
Thirdly in the U.S., political disfunction allows unbridled spending to pump up demand without contributing to the supply side of the economy, a condition that cannot persist. A recent poll found that the approval rating for our current Congress is at a record low of 19%, and with good reason. At the time of this recording on October 18, the House has not yet selected a Speaker and much time has been lost in the attempt to resolve the budget by the mid-November deadline. All that remains is for Congress to declare itself a federal disaster area.
Another continuing resolution that maintains the current unsustainable pace of spending growth - and that figured heavily in the recent rating downgrade of US debt - is likely. And of course, another “government shutdown” threat looms. To that point, it is important to understand what a shutdown is, and what it is not. While investors will see the lack of economic data being reported, and a halt in regulatory reviews of certain transactions, most people will not notice any effects. Many essential federal services will continue without interruption including Social Security and Medicare, air traffic control and TSA, the military, the State Department diplomacy operations, the Post Office and law enforcement arms of the FBI, ATF, and others. Most government employees whose paychecks will stop temporarily will all receive back pay. So beyond sensationalized media coverage, it will not likely have any meaningful economic effects. We expect markets to look through the short-term disruption and focus on the longer-term persistence of inflation and policy drag on economic growth.
So for these reasons, we continue to think that the three bears will eat Goldilocks, that believers in a soft landing will be disappointed, that a recession is likely, and that it will be shallow and relatively short.
One of the reasons we think it won’t be long or deep is the lift from AI. Artificial Intelligence and chatbots are two of the hottest topics in media these days, both social and traditional. Generative AI, large language models, machine learning, and natural language processing, these and related technologies have great potential, certainly for innovative benefits in scientific research, healthcare, industrial processes, and as we’ll see in a moment, markets and investment analysis.
Some folks are concerned that there is potential for risks and abuse as well. Recently, I sat down with my colleague Ole Jorgensen, Research Director with our emerging markets affiliate, Global Evolution, of Kolding, Denmark. Ole has spent a lot of time thinking about AI and his firm has already applied it to certain aspects of their investment process. Our discussion was recorded at a recent gathering and Ole’s insights offer valuable perspectives on this fast-growing area.
A DISCUSSION OF ARTIFICIAL INTELLIGENCE WITH OLE JORGENSEN
RS: Ole, great to have you joining our podcast series. Let’s start with how you and the team at Global Evolution use these technologies, and how they have helped us produce results for clients.
Ole Jorgensen: Thanks for the invitation, Rich.
At Global Evolution, we've been working with these technologies for quite a while, and with Conning staff now also, for quite some time. And I think I would group the research outputs into two groups. One is about pattern recognition, which is something that will give us much smarter. much more accurate signals that we can use as part of the investment process. The purpose of that is to generate performance. So that's one direct element, I call that group “valuations.”
The other group is more about insights, which is more qualitative, where you can use other types of AI that are not necessarily about pattern recognition, but more about tapping into the sentiment ,or the perception, of what is communicated in media, central bank communication, earnings calls and many other things.
How those types of more qualitative insights can be tapped into and measured is something new. So now we are sort of able to measure the immeasurable, and that's a new part. And that is something that can help us get smarter about what we invest in and have a bit more insight. So that's what I call our “insights” products.
RS: So we hear a lot about algorithms. How do algos learn? How do they improve and enhance traditional analytical methods for investment analysts?
OJ: The old analysis we all grew up with was basically like a regression diagram with dots and then a straight line through it. That straight line is way too simple. It's interesting because it gives us easier insights. It's easy to understand what comes out of these linear, old-fashioned models, but they're not very accurate. What the new technologies can give us, especially machine learning, is a very accurate signal, but the problem here is that it's not very transparent. It's like a black box, but they are super accurate and the way that the algorithms actually become accurate is that they try to imitate how our brains search for patterns. That's why the technologies typically are called neural nets.
The neurons in these models, or the artificial neurons, search for patterns between hidden layers that people can't see, but the algorithm can see that. For example, in ChatGPT there are about 96 hidden layers. So that certainly is going deep and very much about deep learning. So the way the algorithms become good is to imitate how humans search for patterns, and that is what actually gives us the very accurate correlations and predictions from such models.
RS: One of the things we hear a lot about in the press is that AI can “hallucinate” the credible sounding but totally fabricated results. Whether such results are accidental or intentional disinformation, how should investment analysts vet information produced by AI when anything it returns could be a deep fake?
OJ: That is an excellent question and to be completely honest, I don't think the literature or the technologies are there yet where we can just let the algorithms run away and do their own thing, and then we can just blindly trust it. We cannot blindly trust them yet. Maybe at some point we can, but that will probably take a long time. So in the meantime, what we really need to do is we need to get these accurate signals. or responses, or images or whatever the AI will give us, and we need to use our sound judgment as people. It is not enough yet just to let the AI run by itself. We need to interpret it, vet it and make sure that we understand what's coming out of it. So in the future - probably way into the future - we can rely on such AI output in various forms in a much more automatic way, but we're not there yet.
RS: I know you're very focused on “sound judgment by people.” To follow up on that comment, people worry a lot that these innovations will kill jobs. That's often the complaint about new and disruptive technologies. Do you see that as a credible threat here? And how should we think about (the job-killing fear)?
OJ: I think that is a very warranted concern, and yes, it will kill jobs, but it will also give birth to new types of jobs, like any other new technology we've seen since man started walking the earth. Any new technology gives rise to disruption, gives rise to new skills being demanded and older skills less demanded. So yes, jobs will be killed and new jobs will be born.
And nothing is new about that phenomenon. So that means that we as people need to be a bit adaptable to change in many forms, both privately and as part of our work, and some types of work will be redundant. It simply will, just like we've seen many, many times over history. But I'm pretty sure that many new jobs will come up, and that will be super interesting for our kids to be part of.
RS: That sounds great.
The ChatGPT launch was about the fastest rollout of any technology in history. What made that particular tool so attractive and easy to adopt and what risks do you think there might be in using a technology that's had such very little time to “shake out the bugs,” so to speak?
OJ: There are a couple of different aspects that I think are very attractive, especially for guys like us in our business, and that is, as part of let's say more back-office or operational functions, we can use these new tools to make some of our work much more efficiently done. So if you're in sales marketing operations, legal, compliance, many, many, many business functions will be sort of “on steroids” with these new language models, simply because you can have like an assistant that can help you out very efficiently and you can use these new technologies to compose and check and write many things very, very quickly, so you will be super performing instead of just performing at regular levels. That's more in the back-office, middle-office element.
In the front office or as part of the investment process, as I mentioned earlier, some of these new technologies can be used to extract information, search for patterns, see through to a large extent those patterns, and give signals for the investment process. And that is for generating performance. So I would categorize the two effects of the new technologies into two groups, efficiency gains and performance gains.
RS: Insurance regulators have been ramping up their capabilities and understanding of AI and other InsureTech methods. For example, the NAIC in the U.S. has formed its Innovation, Technology and Cybersecurity Committee, whose sweep includes AI. A big part of their focus is data security and privacy, as well as anti-bias and anti-discrimination in underlying algorithms. What can you tell our clients to help them understand how we at Conning cope with and mitigate those risks in our work?
OJ: Yes, and it goes a little bit back to what I mentioned before about, you could call it a like a human overlay. We cannot just apply these methods yet without double-checking things. We cannot just blindly deploy these new algorithms, and the reason is it's still a little bit of a black box. So until we are able to understand what's going on in that black box, we cannot just let it operate and make decisions for us. That is one step too far still, so therefore what we will do and should do going forward, also in Conning, is to experiment with these new technologies, use them as input to our decision making processes, but not blindly let them steer anything. Not yet.
RS: So for the last question. How intelligent are these artificial intelligence programs? Can they pass the so-called Turing test, that if you're speaking to it, you can't tell if it's a person or not?
OJ: That's a good question. And actually it seems like if you open or think about opening an AI tool, if we are not that used to it yet, then one would expect something magical to happen just from launching the tool or opening the tool. But once you (open the tool) - nothing happens. It only happens when you give it an order or demand it to do something, or if you tell it “Please give me this” or “Please write something about that,” “Please estimate this,” “Please forecast that.” You may have to make the decisions, you have to decide what's important, you have to be knowledgeable about what the right questions are. If you ask it the wrong question, it's going to give you a response because it doesn't know better yet.
So we still have a job, but the job will be to be knowledgeable, to be aware of what we think is important for our business, and that is then what we can feed into the AI. The AI cannot do anything on its own. We have to tell it what to do. So we're the bosses and they work for us.
CLOSE
RS: Encouraging way to close. Thank you, my friend, for your time today. I really enjoyed hearing you share your expertise and insights with our podcast listeners. And thanks to you and your team for your innovative work and bringing the value of these technologies to our clients.
So, as often happens throughout human history, we must temper our enthusiasm for the potential of exciting new technologies with sorrow for the victims of political terrorism and the fervent hope for its eradication globally. But it is only through vibrant markets and a strong global economy that the forces of good can prevail and the lives so damaged can rebuild.
Thanks for listening in on our discussion. That closes out our podcast series for this year. We hope you’ll join us early in 2024 for our economic and election outlooks.
We hope this information is helpful and we always welcome your feedback and questions. You can get messages to us through your relationship manager, your portfolio manager, or send them directly to us at ConnText@Conning.com (that’s ConnText w/ 2 Ns).
All investment performance information included in this podcast is historical. Past performance is not a guarantee of future results. Any tax-related information contained in this podcast is for informational purposes only and should not be considered tax advice. You should consult a tax professional with any questions.
For complete details regarding Conning and its services in the U.S., you should refer to our Form ADV Part 2, which may be obtained by calling us.
Legal Disclaimer
©2023 Conning, Inc. This podcast and the software described within are copyrighted with all rights reserved. No part of this podcast may be distributed, reproduced, transcribed, transmitted, stored in an electronic retrieval system, or translated into any language in any form by any means without the prior written permission of Conning. Conning does not make any warranties, express or implied, in this podcast. In no event shall Conning be liable for damages of any kind arising out of the use of this podcast or the information contained within it. This podcast is not intended to be complete, and we do not guarantee its accuracy. Any opinion expressed in this podcast is subject to change at any time without notice.
This podcast contains information that is confidential or proprietary to Conning (or their direct or indirect subsidiaries). By accepting this podcast you agree that: (1) if there is any pre-existing contract containing disclosure and use restrictions between your company and Conning, you and your company will use the information in this podcast in reliance on and subject to the terms of any such pre-existing contract; or (2) if there is no contractual relationship between you and your company and Conning, you and your company agree to protect the information in this podcast and not to reproduce, disclose or use the information in any way, except as may be required by law.
ADVISE®, FIRM®, and GEMS® are registered trademarks in the U.S. of Conning, Inc. Copyright 1990-2023 Conning, Inc. All rights reserved. ADVISE®, FIRM®, and GEMS® are proprietary software published and owned by Conning, Inc.
This podcast is for informational purposes only and should not be interpreted as an offer to sell, or a solicitation or recommendation of an offer to buy any security, product or service, or retain Conning for investment advisory services. This discussion may include predictions, estimates, or other information that might be considered forward-looking. While this content represents our current judgment on what the future may hold, it is not a guarantee and the investment market is subject to risks and uncertainties that could cause actual results to differ materially. You are cautioned not to place undue reliance on the content of this podcast, which reflects our opinions only as of the date of this presentation. Nothing contained herein is meant to be investment advice.