How ChatGPT will change the world -and financial advice? #writtenbyahuman

Markets didn’t move much last week, as investors mull over the possibility that the present rally may conflict with the harsh realities of rising interest rates and a swiftly decelerating economy. However, one event stood out. Google’s stock lost 10%, nearly $100 Billion in market cap, after its Bard AI made an error.

So, in a departure from our usual Monday post-mortem and prognosis, we want to address the elephant in the room of the last two months: ChatGPT.

For those who have not heard of it yet, ChatGPT is the world’s first successful AI-driven chatbot. When one Googles a term, they get back a list of articles on the internet. While search engines have become a lot better at personalising and accurately pinpointing information over the past few years, their basic architecture has remained the same since the days of Yahoo! in the mid-nineties.

ChatGPT is revolutionising the ‘answer’. No longer a list of things, the chatbot actually synthesizes original natural-language answers to real questions -even poorly phrased ones. The bot, launched last November, has gained over 100 million users, making it the fastest growing app in history. Just last week, ‘ChatGPT’ appeared in 18,000 articles.

Critics will say that the level of English and discussion is that of a high schooler. For the most part they are right. It presents information in a simplistic ‘first-it-was-the-dinosaurs’ fashion, which is why ChatGPT is initially heavily used by students to find a way around the usual copy-paste searches schools and university teachers run. It is though, still a big leap in technology. Until a few years ago, the world’s most successful chatbot had the intelligence of a 4-year old. Microsoft’s experiment with Tay, a chatbot let loose on Twitter, was an epic disaster. The bot turned into an internet troll and espoused right-wing fundamentalism within the 24 hours of life before its termination. The leap in just six years is enormous.

To test it, I asked ChatGPT to synthesize an original 1,000 word story, based on a short story I once wrote (which is how I knew it was original). It came up with a decent story. Then I asked it to rewrite the story both in an Edgar Allan Poe manner and then as Shakespeare sonnet.

“Eternal Remembrance: A Sonnet for the Departed”

In Memory of the Passed
The stars do rise and shine in the sky
As moments in memory, that do nigh
In memory of the passed, who lived before
Our time on this earth, and are now no more.

Where are they now? Where have they gone?
Are they in God’s love, or in shadow unknown?
In their memory, we live with hope and love
In memory of the passed, sent from above.

Written by ChatGPT

That’s pretty impressive.

ChatGPT can write original viral social media posts, essays, books and virtually any content in a matter of minutes. It can also write code. And don’t think that the Bot doesn’t have an opinion. Asked whether I should vote Republican or Democrat it presented me with a balanced argument. When I identified myself as a liberal, the opinion remained balanced but leaned Democrat as the more likely choice. A decision-making model is the ultimate goal of most AIs out there, so current technology should very easily come up with an eloquent way for the model to convince me to vote for one party.

It is eloquence which brings us to the next issue, ‘Anthropomorphism’. Anthropomorphism is the interpretation of nonhuman things in terms of human characteristics. It is a natural instinct, and the reason why we see a ‘face’ on the surface of Mars instead of random formation. Or the reason why we prefer puppies over geckos, and geckos over spiders as pets (well, most people do anyway). The fact that the language is ‘normal’ is what makes it so attractive. We like talking to something human-esque. The Anglosaxon culture especially gives great importance to eloquence, as a sign of intelligence. It is on this principle that many search engines are built, to identify original content.

Anthropomorphism drives technology. Alexa and Siri are algorithms with human names pretending to be helpers with human voices. In science fiction, the ultimate goal is an anthropomorphic robot. Today’s scientists grew up with Isaac Asimov’s ‘I Robot’ and Ridley Scott’s ‘Blade Runner’, featuring near-human robots. And while trash-can looking Star Wars’ R2-D2 may be lovable, it was 2013’s ‘Her’, where Joaquin Phoenix falls in love with the human voice of an AI that earned the Oscar. In just two months, ChatGPT is changing the landscape for the tech industry. Currently, the search engine world is dominated by Google, with a whopping 93% market share.

Google and Microsoft (Bing) have already announced that they are entering the AI race, as they see the industry-changing disruption wrought by the chatbot. Google, one of the biggest companies in the world, made $162 billion last year from search engine services alone (which account for 60% of its revenue). This implies a search engine industry size of nearly $175 bn. However, the implications are much bigger, as Google represents the main advertising channel of our times. A doctor advertising on Google may pay $200 per month for targeted adverts and make thousands more on the back of that. Sundar Pijay, Google’s CEO, was excited that the search engine industry will get a much-needed revamp and introduced us to Google’s chat AI, Bard. Investors may have been disappointed by Bard’s response, as the AI made an error, and plunged Alphabet’s stock 10% last week, but we are still at early development stages. Humans make mistakes. So do search engines and Google Maps.

Microsoft follows, saying it will add an AI chatbot feature to Bing and incorporate more AI in the myriads of services it sells.  If Google changes the model, or if Microsoft comes up with a better engine, it will have a domino effect on millions of enterprises -and marketing departments.

Chat GPT is already changing the world. It’s threatening to reshuffle the search engine market, and with it change marketing for millions of firms globally. To understand how much and how fast, we need but to travel to 1990, and remember how different a world before search engines was. This is what life before ChatGPT may look to us a few years from now.

The implications of a machine that can write good content, and keeps learning, are momentous. The financial world experimented with Robo-advisers, but the experiment failed, as people would rather interact with humans. Will a human-like AI succeed where dull robots failed? Can algorithms talk to investors in a convincing way? The answer is ‘most probably yes’. And clients will weigh the cost of talking to a human adviser against using a (presumably cheaper) AI.

As humans compete, they will have to come up with increasingly original content. This will eliminate a lot of wasteful content from the financial industry and pre-packaged advice.

However, a Chatbot has some ways to go before being able to provide good advice. A bot will take the input you give it and turn it directly into output. “I want to retire in 10 years”, will say the client, and the Robot will give them an algorithm of how that may happen.

However, good professionals don’t work like that. An insightful adviser will probe to find out not what you want but what you really need. It’s the difference between spam advertisement and personalised recommendations. The new reality will allow advisers to focus on clients who will seek out more meaningful discussions. Good advisers know that, ultimately, best solutions for each client are tailor-made and dynamic.

As for markets? The pure randomness of returns and risks, the element of the unknown as well as policy shifts which are (for the time being) dictated by humans will maintain a healthy enough level of uncertainty for human points of view to still beat markets. Where the smartest algorithms in the world have failed to consistently generate revenue, eloquent bots will probably not succeed. The risk of course is that ‘smart bots’ will convince many human clients to sign up with them, appealing to their andromorphic instincts.

However, much like in pre-GPT life, eloquence is no substitute for efficiency and careful thought. Robo-advisers did not hit a wall because they were not eloquent enough. Their growth stopped because financial advice is a ‘Trust’ profession, and algorithms simply can’t generate trust the way humans do, as many a helpdesk and other simple services are finding out. Financial advice and investment involve complex questions and answers and volatile emotions. AI simply can’t interact that way. If anything, our culture has created a healthy distrust around giving too much power to thinking machines.

Financial advice will, probably, remain human-centric for the time being. But make no mistake, Chat GPT and its offspring will push practitioners away from platitudes and generalised content and towards a deeper understanding of clients and markets.

#writtenbyahuman

David Baker – CIO