Beyond the hype around ChatGPT’s uncanny capabilities, Open AI, the chatbot’s mother company, deserves kudos for attracting such a wide audience, so fast, and with a seemingly insignificant marketing spend. An early investor in Open AI, Microsoft has certainly piggybacked on ChatGPT’s popularity and its perceived innovation edge. Transforming this first-mover advantage to re-position Microsoft Bing as a tangible competitor to Google on search might be a tall order, but Microsoft has the best shot at it that they have had in two decades.
Following the ChatGPT tide that has swept the world since the chatbot launched on 30 November 2022, AI has suddenly become more tangible to a mainstream crowd that long found the very concept of AI rather abstract.
Bringing ChatGPT to the market as the first user-friendly large language model (LLM) tool is no small feat. Nor is stimulating the enthusiasm and creativity of millions of users, in fact millions of voluntary beta testers. Every free user is indeed a convertible paying user, and each new prompt a “free” training instance for ChatGPT.
Microsoft, as an early investor in ChatGPT’s parent company Open AI, had its work cut out to leverage ChatGPT’s popularity and (re)position itself as an AI trailblazer. As it announced the integration of ChatGPT into Bing, Microsoft unexpectedly changed the perception of its ailing search engine to present it as the most advanced search engine on the market. If anything, this is a re-positioning coup that deserves some credit.
Positioning ChatGPT as the coolest generative AI tool around
It would be slightly naive to say that, as a brand, Open AI started from scratch. Famous founders, including serial tech entrepreneur Elon Musk, as well as well-publicized investors including Microsoft, had already made Open AI a brand to watch in the AI tech and research space. The “Elon Musk” brand helped give it technical credibility and a visionary status – all before ChatGPT even got started. Microsoft, who invested USD 1 billion in the company in 2019, added a layer of business legitimacy to the mix.
ChatGPT has now released a solution that tickled the interest of a demanding crowd of early adopters with social clout. A user-friendly interface and well-publicized “wow experiences” generated word of mouth and did the rest. That so many non-technical users were able to put a name on the vague concept of AI certainly helped grow media attention. For the first time, a mainstream audience could start and play, knowingly (few think of Siri or Google Home as AI-driven tools), with a generative AI tool. The “free” media attention ChatGPT has been receiving since its launch is likely worth millions of dollars at this point. And masterminding such a successful viral launch campaign is, no doubt, every digital marketer’s dream.
ChatGPT is one of many LLM tools engaged in the race to lead the market for conversational AI. Other major players include Google, Microsoft, IBM, AWS, Baidu, Oracle, and SAP. That market is expected to grow from USD 6.1 billion in 2021 to 18.4 billion in 2026, according to estimates by Research and Markets. But so far, ChatGPT is the only one that has been made available for free to the general public.
With users’ jailbreaks, correct or incorrect answers to prompts, and new use cases, ChatGPT has succeeded in delivering sticky content that has gone viral. By striking a chord with end-users, from tech professionals to the mainstream public, ChatGPT has in no time received the attention of media outlets and tech influencers around the world –– with little to no marketing costs.
Repositioning Bing as the most advanced search engine on the market
Whether the launch of ChatGPT and Microsoft’s latest USD 10 billion investment in it, together with a new Bing, were part of a branding master plan or not is not only highly speculative but besides the point right now. Answering a question from The Verge, Microsoft’s CEO Satya Nadella said, "brands can be rebuilt as long as there’s innovation,” while rejecting the idea of rebranding Bing.
Microsoft is set to benefit the most from ChatGPT’s rise to fame. To this day, no previous effort has succeeded in making Bing a successful competitor to Google. Google’s technology has been superior, and the company itself consistently more popular than Microsoft in the eyes of end-users and advertisers alike.
Google, on the other hand, has much to lose when Microsoft integrates ChatGPT into Bing and gets the best opportunity in the past two decades to challenge Google’s quasi monopoly. Indeed, most revenues generated by Google come from search, whereas Bing stands for around five percent of Microsoft’s revenue.
By integrating the technology of a brand that the public perceives as visionary, advanced, and cool, Bing has created the best possible playground in which to up its game. Microsoft itself, as a whole, comes out as the biggest winner right now, having its USD 1 billion bet on Open AI pay off. Beyond the financial aspects, the brand upside is potentially huge for Microsoft. The perception of ChatGPT as visionary, inspiring, innovative, and even chicky can also rub off onto the more conservative Microsoft. Open AI’s ChatGPT might just be Microsoft’s ultimate brand ingredient. ChatGPT has undeniably succeeded in taking a position (despite launching after Meta’s BlenderBot) and Microsoft has every possibility to capitalize on the ChatGPT brand to reposition both Bing as advanced and user-friendly, and Microsoft as visionary and cool.
With Google about to release ChatGPT’s competitor Bard, a new search engine war is on the horizon. It is now up to Microsoft to show that they can grow this first-mover position into a long-term advantage. As Al Ries and Jack Trout state in their reference book “Positioning: The battle for your mind”, “The power of the organization is derived from the power of the product, the position that the product owns is in the prospect’s mind. Coca-Cola has power. The Coca-Cola company is merely a reflection of that power. Outside the Coca-Cola field, the Coca-Cola Company has to earn its power the hard way – either by getting into the mind first, or by establishing a strong alternative position, or by repositioning the leader.” Should Microsoft succeed in transforming the short-term momentum into long-term value, they could become a textbook example of Ries’ and Trout’s theory.
Just as well-publicized mistakes show, while chatbot technology may have taken a major leap forward, it still needs to be perfected. All typical flaws of large language models, including biases and hallucinations (when an AI makes up information), are just as likely to be present in Microsoft’s new products as in Google’s. Ironically, Google’s share price took a USD 100 billion hit when Bard committed a mistake while being demonstrated publicly for the first time. Users, the press, and the stock market have in the meantime shown more leniency towards ChatGPT’s biases and hallucinations.
Despite its shortcomings, ChatGPT is the fastest growing publicly-available digital platform to this day. A week after launch, it had already gained one million users. One month after launch, that figure had grown to a whopping 57 million and, by January 2023, ChatGPT had 590 million monthly visits. It had surpassed 100 million monthly active users by the end of January, according to an estimate by investment bank UBS. In comparison, it took TikTok about nine months to build an equivalent user base, and about two and a half years for Instagram to acquire 100 million monthly active users, according to CBS. In a slightly more comparable reference, it took Google Translate six and a half years to reach that level of monthly active users.
If Open AI plays its cards right, it could well succeed in generating revenues of USD 1 billion by 2024, as it stated was its goal. Hundreds of online groups have been created to share knowledge, tips, and tricks about ChatGPT, what it can and cannot do, and what it could do beyond typical chatbot capabilities, such as finding answers to search queries. Tom Goldstein, Associate professor in computer science at the university of Maryland, has estimated the monthly cost of running ChatGPT to be USD 3 million, based on the costs for Azure Cloud, the server infrastructure on which ChatGPT runs. ChatGPT has released a monthly plan at USD 20 per month, and some professional users have been given access to a USD 42 dollar/month Pro plan. The ChatGPT API will also be made available alongside other Open AI APIs. In other words, it looks as though Open AI’s monetization goals are well within reach. This is certainly good news for Microsoft, who has committed a new multi-year, multi-billion-dollar investment in Open AI.
At this point, the reputational risk of getting it wrong seems to be larger for Google than it is for Microsoft, possibly because people have learned to rely on Google for search, less so for Bing. To be fair, ChatGPT makes stuff up because users try to make it perform tasks it was not designed for, including search. The technology Microsoft is deploying in Bing is in fact a more advanced version of ChatGPT called Prometheus, which was designed for search.
Microsoft’s earlier chatbot experiment with Tay, a Twitter chatbot that was coached into posting racist slurs, was retired after a day. Google’s ChatGPT killer Bard managed to return erroneous answers during its launch event. The stakes are high for Microsoft to get it right with Bing. Given that Microsoft has exclusive access to ChatGPT’s base code, is also integrating it in its Office Suite, and has secured the exclusivity for Microsoft Azure to drive ChatGPT, Microsoft could end up winning in more ways than one -- provided ChatGPT becomes more reliable, and that Open AI retains and grows its paying user base. As Ries and Trout write, there is one way to get to be the leader: “You get there firstest with the mostest.”
First-movers are left to playing catch-up
Meta’s release to the public (US only) of its own LLM, BlenderBot 3, back in August 2022, went pretty much unnoticed, despite being the first publicly-available 175 billion machine learning (ML) parameter-chatbot ever launched. A few days after launching, it collected a mere 70 000 conversations and managed to alienate the tech press due to its shortcomings. “There are a lot of different ways to measure the performance of a chatbot. By nearly any of them, Blenderbot is really, really bad,” stated Vox. A key feature of BlenderBot is that it can search the internet and talk about specific topics, that is what one would expect from a search engine with an AI-chatbot interface. BlenderBot 3 can access live data and even cite its sources, unlike ChatGPT which operates on non-real time training data and does not reference sources in its answers.
Google has also developed an LLM called Lamda, short for Language Model for Dialogue Applications. After referring to ChatGPT as code red earlier this year, Google has now announced the release of Bard, a ChatGPT competitor using Lamda, to “trusted testers”. Lamda is built on Transformer, a neural network architecture created by Google Research in 2017. Ironically, Transformer is also the foundation for ChatGPT. It launched its own AI chatbot, A, last May, which today has about one million users. Open AI has an undeniable first-mover advantage here, not only when it comes to launching first, but also to have trained ChatGPT on such a large data set. This creates a situation where competitors are left to play catch up. Nothing says that a new jack-in-the-box generative AI tool couldn’t pop up and challenge ChatGPT. Asian telecom operator SK Telecom, which repositioned itself as an AI company in 2022, is one of the contenders. SK Telecom already launched its own AI-powered chatbot last May. Called A. (pronounced "A dot"), it is similar to ChatGPT but in Korean. The service has attracted over 1 million subscribers domestically, the company said at an MWC event.
In need of some business building insights? We are brand and business design. Let’s talk.
- Generative AI: Generative artificial intelligence (AI) describes algorithms used to create new content, including audio, code, images, text, simulations, and videos. Source: McKinsey
- Conversational AI: Conversational AI describes technologies, such as chatbots or virtual agents, which users can talk to. To help imitate human interactions, recognizing speech and text inputs and translating their meanings across various languages, these technologies use large volumes of data, machine learning, and natural language processing. Source: IBM
- Large language model (LLM): Large Language Models describe artificial intelligence tools that can read, summarize and translate texts as well as predict future words in a sentence, which lets them generate sentences that are similar to how humans talk and write. Source: University of Michigan
- A machine learning parameter is “a configuration variable that is internal to the model and whose value can be estimated from the given data.” Source: Datacamp
References, articles, and further reading
Positioning: The battle for your mind, Al Ries, Jack Trout, McGraw-Hill, 2001
New jailbreak! Proudly unveiling the tried and tested DAN 5.0 - it actually works - Returning to DAN, and assessing its limitations and capabilities.
Google CEO Confirms AI Features Coming To Search “Soon”
OpenAI is giving Microsoft exclusive access to its GPT-3 language model
Microsoft's investment into ChatGPT's creator may be the smartest $1 billion ever spent
How ChatGPT exploded onto the scene with so little marketing spend
BlenderBot 3: A 175B parameter, publicly available chatbot that improves its skills and safety over time
BlenderBot 3: a deployed conversational agent that continually learns to responsibly engage
Deep Reinforcement Learning from Human Preferences (Open AI)
Gartner’s magic quadrant for enterprise conversational AI platforms 2022
Google shares drop $100 billion after its new AI chatbot makes a mistake
A South Korean telco giant has its own A.I. chatbot — and says it’s a ‘super app’ version of ChatGPT
SKT doubles supercomputer capacity to speed up launch of Korean ChatGPT