Artificial intelligence-powered ChatGPT arrived in a firestorm nine weeks ago, wowing the tech world with its ability to do everything from composing a high school essay to debugging complex computer code. Investors quickly fastened on how the groundbreaking content-creation tool from startup OpenAI posed a threat to Google stock and the search giant’s market share.
But ChatGPT is much more than just a peril to Google and its parent company Alphabet (GOOGL). And it’s only one of many “generative AI” technologies that could roil a host of industries by creating text, images, video and computer programming code on their own.
The technology quickly captured the attention of tech giants. One was Microsoft (MSFT), which doubled down on its early investment in the company. Microsoft said Jan. 23 that it’s making a “multiyear, multibillion-dollar investment” in OpenAI. It invested $1 billion in OpenAI in 2019.
OpenAI’s Nov. 30 release of ChatGPT to the public marked the first time such a powerful AI tool was at the disposal of ordinary internet users. Yet generative AI technology already is finding applications in marketing, advertising, drug development, legal contracts, video gaming, customer support and digital art.
Disruption Seen Coming
“It feels like a moment where there’s a lot of disruption coming,” said Sridhar Ramaswamy, co-founder and chief executive of internet search startup Neeva and a former Google senior vice president. Ramaswamy oversaw Google’s digital advertising business for five years.
ChatGPT uses AI to produce humanlike responses to search queries based on data scooped up from the internet. Ask it to solve a complex math question and it’s done in seconds. You can get instant translation into English of text in a foreign language. Or give it an inventory of what’s in your refrigerator and it can produce more than a dozen recipes from those contents. The list is endless.
And it’s no wonder that within a few days of its release, ChatGPT had more than 1 million users.
To be sure, the impact of artificial intelligence on human society and the business world has been overhyped before. Other potential buzz-killers include huge cloud computing costs associated with ChatGPT. OpenAI uses Microsoft’s cloud computing arm to process search requests.
Further, ChatGPT is not updated in real time through internet connections. It has limited knowledge about things that occurred after 2021, when its training period ended. OpenAI also cautions that ChatGPT is not perfect, particularly for students planning to use it to cheat on schoolwork.
Meanwhile, intellectual property issues arise since generative AI models scrape the internet for content they need to create new material. And generative AI technology could be used for nefarious purposes, say, “deepfakes” in video or helping computer hackers generate code. At the least, OpenAI concedes that ChatGPT sometimes generates incorrect or biased content.
Still, the frenzy over ChatGPT is real. Early users experimented with ChatGPT, asking it serious and not-so-serious questions, probing for strengths and weaknesses.
In its first month, ChatGPT scored nearly half the unique visitors that Microsoft’s Bing search engine drew.
OpenAI’s computing resources hit a wall following the explosion in usage. By mid-January, OpenAI said ChatGPT was “at capacity” and connections to its cloud system were at times unavailable.
Add up the pros and cons, though, and generative AI looks like it will make its mark. And Google stock, among others, is in its path.
Artificial Intelligence: Natural Language Models
“Internet search really has not been innovated on since the pay-per-click model that Google pioneered almost 20 years ago,” Ramaswamy told Investor’s Business Daily. “So the business model is ripe for disruption. I think we’re on the threshold of massive changes in how search is going to operate and how AI is going to produce better experiences for users.”
While Neeva’s search engine is ad-free, it charges a $5.99 monthly subscription for access. It’s in the early stages of using generative AI.
Key to the rise of generative AI are improved natural processing language models that help computers understand the way that humans write and speak. OpenAI is part of a wave of NPL startups that includes AI21 Labs, Anthropic, Cohere and others.
Google and Anthropic on Feb. 3 announced a partnership. As part of the deal, Google invested $300 million in Anthropic, said a report. Google shares were down 2.5% Friday, largely due to the company’s earnings results from Thursday.
Large language models provide the building blocks to develop applications. These models process “prompts,” such as an internet search query, that describe what a user wants to get.
Microsoft plans to integrate AI tools like ChatGPT into all of its products, including its Bing search engine.
AI Startups Leverage Generative AI
Microsoft has already integrated OpenAI’s Dall-E2 tools into its graphics design software. Dall-E2 converts written text into pictures. OpenAI’s generative AI could also improve email platform Outlook, presentation tools and office productivity tools such as Word.
Many technology companies are racing to integrate generative AI into products.
Austin, Texas-based Jasper, a marketing software company, pays OpenAI for access to GPT3, its older language model. Jasper’s marketing department customers describe in natural language what they want Jasper to write.
Jasper’s AI platform generates content for blog articles, social media posts, website copy and more. It recently raised $125 million at a price valuing the company at $1.5 billion.
Meanwhile, startup companies including Adept, Inflection AI, Character.AI and others aim to automate software tasks and integrate generative AI into third-party applications.
ChatGPT: Enterprise Software Transformed?
“I think that the rise of generative AI has the potential to be as impactful, if not way more impactful, to enterprise software as cloud computing and SaaS (software-as-a-service),” Zukin told IBD. “Microsoft is betting the farm on OpenAI, and we think this is definitely just the tip of the iceberg.”
Generative AI technology is finding applications in legal contracts, copywriting, video gaming, customer support, computer programming and digital art. Marketing departments are among the early adopters, says Thomas Davenport, a Babson College professor.
“For creative work there are going to be companies that adopt it in a transformative way,” said Davenport, co-author of “All-in on AI: How Smart Companies Win Big With Artificial Intelligence.”
He added, “A lot of companies are doing generative AI models. Some are open-sourced, while some are proprietary. It’s changing very rapidly.”
Venture Capital Jumps Into Generative AI
Some builders of large language models, such as Cohere and AI21 Labs, sell their own products in the enterprise market. Others make their models freely available to software developers.
Aside from OpenAI, another well-funded startup is Stability AI, the developer of open-source text-to-image generator Stable Diffusion.
Startups targeting fashion, e-commerce, drug design, personalized videos and computer vision include Veesual AI, Runway, Ordaos Bio, Paige.AI, Synthesis AI, Character.AI and Tavus.
According to PitchBook, venture capital investment in generative AI jumped to $1.37 billion in 2022 — almost as much as was invested in all the previous five years combined.
Big Pharma Targets Drug Discovery
Research firm Gartner forecasts that generative AI will be used to discover 30% of new drugs and industrial materials by 2025, up from zero last year.
Gartner analyst Brian Burke told IBD that drug companies are using generative AI to design properties or functions of protein models that target diseases.
“Virtually all the large pharma and lots of small pharma startups are working on generative AI,” he said. “It’s been in development for a couple of years. Some drugs are in clinical trials now. It’s going to be a major shift in the pharma industry.”
How Did Your Online Broker Do In IBD’s 2023 Best Online Broker Survey?
Generative AI requires massive computing power — data-intensive capabilities such as machine learning and neural networks to analyze patterns in text, images, voice recordings and other digital sources. Neural networks are modeled on the human brain.
Machine learning systems analyze huge troves of data to train the software algorithms that underlie large language models. Deep-learning AI tools fine tune algorithms for specific tasks, such as generating realistic human text.
SambaNova Backers Include Intel, Google
A fourth-generation version of OpenAI’s large language model, GPT-4, is expected to be released in 2023. OpenAI has stated that the foundational data of GPT-4 won’t be much bigger than GPT-3, which was trained on 175 billion “parameters.” But GPT-4 will use more computing power.
Meanwhile, chipmaker Nvidia (NVDA) provides software development tools to build artificial intelligence applications. Nvidia also has developed its own large language models. Earnings for NVDA stock are due Feb. 22.
In November, Nvidia agreed to partner with Microsoft to build a new cloud computing supercomputer to process AI workloads.
One Nvidia chip rival, well-funded startup SambaNova, is targeting the banking, energy and health care industries. Palo Alto-based SambaNova’s investors include BlackRock, Intel‘s (INTC) Intel Capital unit and Alphabet venture arm GV, formerly Google Ventures. It builds pretrained, custom models using OpenAI’s GPT-3.
SambaNova’s biggest customers are U.S. government research labs. It’s building an AI supercomputer with Hungary’s OTP Bank.
“We are now as a world moving from thinking about what the possibilities are in using AI to productivity,” Chief Executive Rodrigo Liang told IBD. “We are taking ChatGPT and these large language models and, in banking and other applications, turning them into useful things that save businesses money, that engage customers better.”
Meanwhile, Babson professor Davenport says generative AI could finally help push AI technology into the mainstream in the enterprise market. Most companies use AI tools to spot patterns and make predictions for apps such as fraud detection and energy exploration.
ChatGPT: Existential Threat To Google Stock?
“Aggressive AI adoption has gone more slowly than some people may have thought,” he said. “A lot of companies have experimented but (the number of) those having large numbers of deployed applications is still quite small. It’s not easy.”
Cloud computing companies such as Amazon Web Services, part of Amazon.com (AMZN), and Google sell AI analytical services to business customers. That makes it easier for companies to bootstrap AI applications if they do not have enough internal resources to train models on foundational data.
Cloud computing companies are anxious to leverage generative AI’s potential. Google’s cloud computing arm has partnered with Cohere to offer AI services. Stability AI and search engine startup Neeva both use Amazon Web Services.
Amid the sudden rise of the OpenAI/Microsoft partnership, the hot topic on Wall Street is what it means for Google stock. Google garners about 57% of its revenue from search products. Google controls 75% of the paid search market while Microsoft’s Bing has 5%, said a Barclay’s report.
ChatGPT Providing The Answer
ChatGPT poses a threat to Google stock for some queries, D.A. Davidson analyst Gil Luria said in a note.
“Google helps you find an answer; ChatGPT provides the answer,” the Google stock analyst said. As critics point out, though, ChatGPT sometimes provides wrong answers or biased results.
Meanwhile, Google-parent Alphabet uses artificial intelligence in its core business and long-range projects such as its autonomous vehicle unit Waymo. Google also uses AI to deliver web search results and relevant digital advertising.
In a blog, Google recently touted its AI innovations. “We expect a launch of new AI prototypes, and products could be showcased as early as the next major Google IO event in May 2023,” said a Bank of America report.
Also, Google owns DeepMind Technologies, a U.K.-based AI research laboratory. Alphabet acquired DeepMind for $500 million in 2014.
According to Bank of America, DeepMind had about $1.05 billion in revenue in 2020 and turned profitable that year. DeepMind in 2021 spun off a new Alphabet venture called Isomorphic Laboratories, dedicated to drug discovery.
Will Google Risk Adding Chatbots To Search?
In addition, Google has developed generative AI tools called Language Model for Dialog Applications, or LaMDA, which use technology similar to OpenAI’s. LaMDA is designed for dialogue while ChatGPT is text-based.
“On the AI side, it is a really exciting time,” Google Chief Executive Sundar Pichai said on the company’s fourth-quarter earnings call with analysts Thursday.
“I think we’ve been investing for a while, and it’s clear that the market is ready. Consumers are interested in trying out new experiences,” Pichai said. “I think I feel comfortable with all the investments we’ve made in making sure we can develop AI responsibly. And we’ll be careful.”
Google plans to make its own large language models, such as LaMDA, publicly available. Google says it will provide tools for software developers, content creators and partners that enable them to build artificial intelligence applications.
Pichai said consumer products, such as conversational AI enhancements to search, will be rolled out in the “coming weeks and months.”
Heavy Investment From Google
“Google has invested heavily in machine learning and artificial intelligence,” said Google stock analyst Mark Mahaney from Evercore ISI. “It has a ChatGPT application that is just as good if not better. I could be wrong — maybe this is the beginning of the end for Google, but I’m skeptical of that.”
Despite the buzz about OpenAI and Microsoft, Google stock has advanced nearly 20% in 2023, ahead of all the major stock averages.
According to one report, Google now plans to embed chatbot features into its search engine. But Google cannot launch flawed products that use generative AI, said a UBS report on ChatGPT.
“Google cannot afford to sully its search results with error prone, biased or toxic results that can occur with these models,” UBS analyst Lloyd Walmsley said in the note.
“Search is Google’s core product and moneymaker. Smaller players like Bing, DuckDuckGo and Neeva can take risks that Google would traditionally avoid,” the Google stock analyst went on to say. “This puts Google at risk to decide between either losing query share or rushing a competing product to market that could reduce quality.”
Another view is that putting chatbot features into its search products will lower Google’s profit margins because of the computing required. Morgan Stanley analyst Brian Nowak estimated in a report that ChatGPT’s cost per query is about seven times higher than Google’s search costs.
Can OpenAI Generate Revenue, Or Just Content?
Still, Nowak expects generative AI to take off.
“We believe AI tools will be key to driving the next generation of consumer and enterprise apps and tools somewhat similar to the mobile ecosystem 10-plus years ago,” he said in a report.
Still, OpenAI’s economics could be a hurdle. Some analysts estimate OpenAI’s annual costs to run ChatGPT on Microsoft’s cloud computing infrastructure at $1 billion. “The compute costs are eye-watering,” OpenAI Chief Executive Sam Altman once posted on Twitter.
With its $1 billion investment in 2019, Microsoft snatched OpenAI away from Google as a cloud customer. OpenAI agreed to use Microsoft’s Azure instead. Microsoft recently issued weak guidance for its cloud computing unit.
Microsoft also gained exclusive rights to some underlying AI technology. Meanwhile, Microsoft’s 2023 investment could give it a 49% stake in OpenAI, according to reports. Microsoft did not disclose the financial terms of the OpenAI partnership.
OpenAI generates most revenue currently by making advanced models available to developers. The company has been testing ChatGPT Plus, a chatbot version for business users. In trials, OpenAI charges $20 monthly for ChatGPT Plus.
OpenAI aims to generate $200 million in revenue in 2023 and $1 billion in 2024.
Follow Reinhardt Krause on Twitter @reinhardtk_tech for updates on 5G wireless, artificial intelligence, cybersecurity and cloud computing.
YOU MAY ALSO LIKE: