February 9, 2023
I am obsessed with generative AI — the newest iteration of artificial intelligence. I am not a technology genius. I have made a living for over 50 years by identifying new emerging technology. In the early days I started magazines, then seminars and trade shows. Now I write this blog and find emerging technology to get excited about and to invest in.
I’ve written this blog for over 20 years. This is the first time I’ve dated it at the very beginning. I want my children and their children to read it and compare my predictions today with what actually happened when I’m long gone — which will be soon. In June I’ll be 81 — way past when my parents and their parents died.
Unless you’ve been vacationing on Mars for the past few years, everyone here on Earth has gone bonkers over generative AI in general and ChatGPT in particular. Similarweb.com reports that on January 31, 2023, the ChatGPT website was visited 28 million times, up 165% from a month previously. No other new technology has experienced this level of interest this quickly.
No other technology!
How do I explain generative AI / ChatGPT to myself?
+ It writes code.
+ It writes blog posts. (Replace me? Egads!)
+ It writes legal briefs.
+ It writes poems.
+ It writes songs.
+ It writes recipes.
+ It writes papers. Soon books.
Late last night I asked ChatGPT some questions:
Harry: Which public companies are likely to benefit the most from ChatGPT?
ChatGPT: As a language model, I don’t have personal opinions or the ability to predict the financial performance of individual companies. However, companies that focus on natural language processing, artificial intelligence, and virtual assistants may benefit from the advancement and development of language models like ChatGPT. This includes technology companies like Google, Amazon, Facebook, Microsoft, and IBM, as well as start-ups that are working on these cutting-edge technologies.
Another area where language models can be applied is in customer service, where virtual assistants powered by AI and NLP can handle customer queries more efficiently and effectively than human agents. Companies in industries such as e-commerce, telecommunications, and finance that have large customer service operations may benefit from using virtual assistants powered by language models.
Harry: How could a bank use ChatGPT?
ChatGPT: Banks could use ChatGPT in several ways to enhance their customer service and improve their operations. Some of the ways are:
- Virtual Customer Service Assistant: ChatGPT can be integrated into a bank’s customer service platform as a virtual assistant to answer customer queries and provide information about products and services. This can reduce wait times and improve the overall customer experience.
- Fraud Detection: ChatGPT can be trained on large datasets of customer interactions to identify potential fraud and suspicious activities. This can help banks to proactively detect and prevent financial crimes, improving the security of their customers’ accounts.
- Risk Assessment: ChatGPT can be used to analyze large amounts of customer data to identify potential risks and improve credit risk assessment. This can help banks to make more informed lending decisions and reduce the risk of loan defaults.
- Chatbots for Lead Generation: Banks can use ChatGPT to create chatbots that engage with potential customers and generate leads for their various financial products and services.
- Back-office Operations: ChatGPT can also be used in back-office operations such as data entry, document processing, and other routine tasks. This can help banks to automate manual processes, reduce errors, and improve efficiency.
Harry: How could a search engine like Google, Bing or Safari use ChatGPT?
Search engines like Google, Bing, and Safari could use ChatGPT in several ways to enhance their user experience and improve their search results:
- Natural Language Processing (NLP): ChatGPT can be used to improve the understanding of natural language queries and provide more accurate search results. This can help search engines to better understand the intent behind a user’s query and provide results that more closely match what the user is looking for.
- Conversational Search: ChatGPT can be integrated into a search engine’s interface as a conversational assistant, allowing users to ask follow-up questions and receive more detailed answers. This can help search engines to provide a more human-like interaction and improve the overall user experience.
- Question Answering: ChatGPT can be trained on large datasets to answer questions directly, providing users with quick answers to common questions without the need to navigate to other websites.
- Recommendations: ChatGPT can be used to recommend content to users based on their search history and other factors, helping search engines to personalize the user experience.
- Content Creation: ChatGPT can be used to generate high-quality content such as summaries, captions, and meta descriptions, which can improve the user experience and help search engines to better understand the content of a page.
If you’re a big company you could use this thing to query all your disparate databases and come with real answers to which products and services are the most profitable, which ones are in most demand (but not being fulfilled) and which you should focus on .
Now what is this thing?
It is super-powerful computer code able to gather intelligence and facts from the vast breadth of the Internet, books and other sources – more information that’s ever been accessible to humanity, digest it, and process our requests quickly into answers and actions we need and can use. Best of all, it can communicate with you in conversational text.
This is Google on steroids. Why search for something and tabulate the results when you can just ask the all-knowing Oracle? It’s like you’re a five-year-old again, and you’re asking your parents a question, perhaps an unexpectedly “deep” one, and expecting them to answer in a way you can understand. They couldn’t always do that. This can. It’s the ultimate know-it-all friend. (In fact, I just read a story about a lady who’s fallen in love — literally — with her chatbot. And she knows it’s a computer. But she prefers it to the recent bad human relationships she’s had.)
Once this thing gets connected to the Internet of Things (IoT), the genie will have left the bottle and you will be able to command it to actually DO things, and it will obey….
— Is this electricity just as it was invented – with no idea what its ultimate uses would be?
— Is this the Internet just as it was invented – with no idea what it would be used for?
— Is this the Internet cloud just as it was invented – with no idea what it would be used for — other than backing up family photos?
When a new invention appears, there is always great, infectious optimism at first.
I spent decades publishing magazines that reported on telecommunications. I know that when the telephone was invented, many naïve people believed that the ability to communicate instantly over long distances would bring people closer together. They thought it could help break down cultural differences. Some people even suggested that the telephone would help to promote world peace. They believed that by fostering greater understanding and communication between people, the telephone would help to reduce conflict and promote cooperation between nations. Alexander Graham Bell himself said the telephone had “the power to unite humanity and foster peace and understanding.”
Well, things didn’t work out that way. Not for the telephone and not for any of the other revolutionary inventions in human history.
Take nuclear energy as another example. It was initially seen as clean, safe and reliable, eliminating the need for fossil fuels. In the 1950s, magazines like Popular Science said that power generated from nuclear plants would be “too cheap to meter.” I kid you not. There were no hints then on the dangers of nuclear radiation and the risks of storing nuclear waste, that ultimately killed its expansion in America and some other countries. But nuclear technology did lead to huge military arsenals that could destroy the world in just an hour or .
Powerful tools are double edged swords. They can be used for both good and evil.
And so today…. Are we now looking at the most powerful knowledge tool ever invented? This seems to have power to do infinite good or to do harm to our thinking and thus what we are likely to do about it? Could this be the beginnings of “insurrection by machine?” Imagine an obsessive super-intelligent machine, without the need for a human at its controls, running amok.
Fathoming the mystery and ethical implications of AI is beyond me. That’s the job of philosophers, scientists and other researchers. Eventually the UN and governments of the world will get deeply involved, and things will really get mucked up.
The readers of my blog (and me) are investors, and are looking for investment opportunities…
So What’s an Investor to Do About Generative AI?
The future of Generative AI for business will be huge, and chaotic.
There is a fork here, between the companies developing Generative AI for themselves or others and those companies that best leverage the resulting technology. Some have already been snapped up by the big guys.
The giants attempting to muscle their way in and control the field are, as you might expect, Alphabet (Google and subsidiaries), Amazon.com, Meta Platforms Inc. (think Facebook, Instagram, etc.) and Microsoft. A wildcard here is the Chinese tech giant Baidu, whose shares surged when it said it will be testing its own generative AI chatbot in March 2023, called “Ernie Bot” (“WenxinYiyan”).Even so, the two principal companies battling over this up in the financial stratosphere are Microsoft and Google.
ChatGPT is the product of a company called OpenAI, which was co-founded in 2015 by Elon Musk and Sam Altman, it’s a research organization focused on developing and promoting friendly AI. OpenAI created GPT-3, one of the largest language models currently available. It led to ChatGPT, which is causing all the hubbub worldwide. The company also developed DALL-E, another amazing generative AI program that can produce realistic images based on a text description you give it, and will eventually put all illustrators out of a job.
Microsoft went bananas over this company, sinking $1 billion into it in 2019, more in 2021, and recently announced a $10 billion multi-year investment. OpenAI uses Microsoft’s Azure cloud exclusively to train all of its AI models. Microsoft’s billions will enable OpenAI to speed up its research and keep it ahead of the competition. Last I read OpenAI has fewer than 400 employees. Microsoft has just added OpenAI’s generative AI technology to enhance the Bing search engine, which has scared the bejesus out of Google.
I tested Bing against Google in the middle of last night. They gave different results. I tended to prefer Google. But it’s early days.
OpenAI isn’t yet publicly traded, but Business Insider has just reported that Shark Tank star Kevin O’Leary told them on a video call a couple days ago that he’s “in talks to make an equity investment in the company, and hopes to solidify a deal within 90 days.”
In the article, O’Leary advises that backing first-movers like OpenAI is a good bet. He says: “AI’s sort of the new, hot kid on the block. Like the internet was 20 years ago. This is the next thing. And what you learn with next things, it’s often best to just invest in the first mover, and sit back and watch.”
Google is panicking over the Microsoft/OpenAI alliance which threatens to usurp Google search. This is particularly ironic and amusing since Google developed the underlying “transformer” AI technology that led to the development of language models such as ChatGPT in the first place. It all started with a 2017 scientific paper by Vaswani et al. entitled, “Attention is All You Need.” The transformer architecture was found to be terrific for natural language processing tasks. Indeed, it’s become the dominant architecture for building large language models (LLMs) and it’s used to build many state-of-the-art systems, most famouslyOpenAI’s GPT series of models, which includes of course the celebrated ChatGPT. Google demonstrated a large language model called Lamda in 2021 but got cold feet when it came to launching a consumer product. Google is now trying to catch up with its announcement of Bard, powered by the same language model, that will compete with ChatGPT.
Beware the Generative AI Bubble.
The problem is, you can use Generative AI everywhere, and everyone is going crazy throwing money around before determining what products and services, if any, are going to be real. Generative AI could be the new Blockchain bubble. (See my blog yesterday for magical thinking.)
A recent Bloomberg headline reads: “ChatGPT Unleashes Stock Trader Stampede for Everything AI.”
As BubbleVision (CNBC) notes: “It (ChatGPT) has been mentioned on earnings calls by management from a range of companies including oil giants, banks — and even the industrial behemoth Caterpillar.”
Another Business Insider headline screams: “ChatGPT has investors scrambling to get on the AI bandwagon so much that obscure stocks are soaring just like big-name chip companies.”
They note that, for example, a little-known staff-recruitment company called Staffing 360 enjoyed a 25% stock jump immediately following news of an AI deal before paring gains.
Any company can now make an AI announcement and boost its stock. Take the lackluster media company BuzzFeed Inc. It got a shot in the arm when its CEO announced he would make AI-generated content part of its core business. Kaboom! Their stock jumped more than 300% over two days last week.
Initially, three broad areas have immediately benefitted from Generative AI:
Customer Service: Automated customer service frees up time for human representatives to handle more complex inquiries.
Content Creation: Businesses are already using generative AI to crank out content for websites, social media, and other platforms.
Marketing: Companies use generative AI to personalize marketing campaigns and create targeted ads. For example, Persado is a marketing technology company that uses generative AI to generate high-converting ad copy. The company has raised over $66M in funding and they work with major brands such as Comcast, Vodafone, Old Navy, GAP, etc. Their revenue is $23.7M and they have a valuation of more than $200M.
And once more research and development occurs, AI in Healthcare will be huge.
This is just scratching the surface. Potentially any company can benefit from the technology, but there will be some companies that focus primarily on the AI aspects, defining their whole “persona,” such as Luka (dba Replika). Currently a privately owned company, it offers a personal AI that uses generative AI to create a unique AI-powered companion for each user. The company has raised over $10.9M in funding over three rounds and has over 5 million users.
Market Insider has reported that C3.ai Inc. stock has soared 86% year-to-date, riding on AI frenzy, especially after it announced that it would integrate ChatGPT into its C3 Generative AI Product Suite. Their suite of AI tools integrates the latest AI stuff from organizations such as Open AI, Google, and the academic world, along with the most advanced language models, such as ChatGPT and GPT-3, into C3 AI’s enterprise AI products.
Then there are companies that indirectly benefit from the AI craze. For example, Nvidia stock has benefitted by Wall Street analysts touting, since its Graphics Processing Unit (GPU) graphics chips are used to compute the AI software models. But Nvidia has grown a bit complacent in recent years and now has some ambitious competitors, principally AMD. Intel is also trying to be an up-and-comer in this space, particularly with their ARC GPUs.
My best advice is to wait a bit and observe the financial AI landscape, and don’t throw money at anything that screams “Generative AI” or what-not. That doesn’t mean that AI won’t have a tremendous future. It will. But since a rising tide lifts all boats to a greater or lesser extent, picking good companies that can most effectively leverage this technology will be, as always, a challenge.
But frankly, as I write this in the nice warmth of a comfy bed in Spencertown, in rural upstate New York, I’m a little afraid. Rumors have been spreading about OpenAI’s up-and-coming, presumably amazing GPT-4.
This would be a really bad time to binge-watch science fiction movies where computers and AI take over the world, like the Matrix, The Terminator, The Invisible Boy, Colossus: the Forbin Project. There are enough things in the world to give you a nightmare.
At 3:00 AM or so I asked ChatGPT to write a poem in the style of Banjo Patterson (famous Australian poet) celebrating Harry Newton as the best tennis player in Columbia County. It obliged. And in less than a minute it penned this masterpiece. Hint: use ChatGPT to write a St Valentines love poem for your partner.
In putting today’s blog together, I had the assistance of the smartest computer whizz I know — Richard Grigonis. Thank you Richard.
See you guys tomorrow. — Harry Newton
