The Benefits and Challenges of Being Second
How Google’s position as second-mover affects their release of an AI chatbot
The common wisdom in business is that the first mover generally wins.
By being first to market, a company builds credibility and gains customers, in turn making it difficult for future competitors to catch up.
It’s the business equivalent of running up the score in the first quarter.
What’s often neglected however is the position of the second-mover.
In the world of generative AI chatbots, the roles of first and second mover have been established by OpenAI and Google, respectively.
Generative AI in 30 seconds
For those out of the tech loop, generative AI is a type of artificial intelligence that can produce unique outputs like text or images, without relying on instructions.
Effectively, it’s AI that creates content on its own.
Recent Developments
The hype around generative AI can be attributed to a host of recent developments in the field. Language models, the computer programs used to interpret and generate text in chatbots, have been super-charged with a new AI development called transformers, which allows them to understand the relationship between words.
Moreover, developments in GPUs, the hardware in computers responsible for calculations, have enabled language models to learn massive datasets.
The combination of transformer and GPU improvements has resulted in generative AI models capable of interpreting and producing complex text and gargantuan quantities of data.
We’re now witnessing generative AI’s gold rush in full form.
The Fight Between Google and OpenAI
OpenAI releases ChatGPT
In December of 2015, a group of AI researchers and prominent tech entrepreneurs founded OpenAI, with the purpose to develop AI tools to empower people.
Since inception the company has released a series of powerful AI tools, including DALL-E, an AI image generator and CodexAI, an AI-powered coding assistant.
In November 2022, OpenAI launched ChatGPT, a chatbot powered by generative AI that can understand user inputs and respond to them.
OpenAI benefited greatly from being a first-mover, as users had never seen a chatbot capable of generating intelligent, almost human-like responses.
In just five days, the app achieved 1 million unique users, becoming the fastest application to do so, and as of January, OpenAI reported that ChatGPT had 100 million monthly unique users.
Google’s less than stellar response
Google’s release of their own AI chatbot has been less momentous.
Soon after ChatGPT’s launch, reports surfaced that Google issued a code red internally.
The reports depicted Google as being threatened by the rapid success of ChatGPT.
The tech giant needed to respond.
In February, four months following the launch of ChatGPT, Google released via waitlist their own beta version of an AI chatbot, Bard.
Despite Google’s status as tech behemoth and leader in AI, the public response to Bard has been underwhelming.
In releasing their chatbot second, Google couldn’t generate the same level of hype for Bard as OpenAI did for ChatGPT. It remains unclear whether Bard’s attained a million users (odds are it hasn’t or Google would have mentioned so in their I/O call)
Google’s Second Mover Advantages
Despite the slow start, Google is far from losing the chatbot race.
The tech behemoth has extensive experience working with language models from developing their advanced model LaMDA
They are also sitting on $100B in cash, which they can mobilize for research and promotion at will.
As a second-mover, Google also has a host of advantages, which it has already begun exploiting.
Second-mover advantages vary depending on the industry, but generally involve the second-mover learning from and adapting around the strategy of the first-mover.
Market Positioning
Google’s primary second-mover advantage is that they can position Bard as an answer to ChatGPT’s inaccuracy.
ChatGPT is notoriously inaccurate.
Ask it to do a basic multiplication problem and the AI will be baffled.
Or better, ask it if there are any countries whose names start with the letter V. It won’t remember Venezuela!
Some of this error stems from ChatGPT’s training data, which goes back to 2019, and ChatGPT’s lack of connection to the internet (as of May 19, 2023).
The overall inaccuracy of ChatGPT suggests the chatbot can be used to spread misinformation. Moreover, it undermines ChatGPT’s use in professional or academic circles.
Almost as if directly responding to OpenAI, Google has oriented their chatbot’s design and brand around accuracy. In line with this goal, they connected Bard to the internet, enabling the chatbot to more accurately answer questions using live data.
They have also dedicated a team of product experts and social scientists to intentionally stress test Bard to probe it for errors and potential harm.
This positioning around accuracy and responsibility is smart for a few reasons:
- It addresses ChatGPT’s inaccuracy
- It aligns with Google’s brand of “organizing the world’s information”
- It explains why Google didn’t launch Bard first
In addressing one of ChatGPT’s greatest weaknesses, Google is subliminally marketing itself as the superior chatbot.
Moreover, this messaging aligns with Google’s overall brand around trust, which emphasizes preserving and spreading knowledge.
It also provides a valid excuse for why Google didn’t release their chatbot first — they were ensuring accuracy!
Why accuracy matters for Google
As regulation around AI takes shape and overall public scrutiny towards misinformation heightens, accuracy will be an increasingly critical factor to be factored into chatbot technology.
Google’s positioning around accuracy now, both in branding and technologically, will help solidify their status as the responsible AI leader.
Second-Mover Advantage #2: Lower Computing Costs
By launching later, Google may also benefit from lower computing costs.
Current generative AI companies have yet to establish profitable business models (as proclaimed by the CEO of OpenAI himself).
This unprofitability is from generative AI’s heavy reliance on data centers and power, which skyrocket computing costs.
One way to lower computing costs is to use more efficient computers.
Of course, easier said than done.
But Google may have an edge.
Last month, they announced their latest supercomputer, TPU v4, and its preliminary results are impressive.
The new computer is 1.2x–1.7x faster than the Nvidia A100 (the datacenter GPU used for OpenAI’s supercomputer).
It also uses 1.3x–1.9x less power.
For Google that translates to cheaper computing costs for running a quicker Bard.
In the long run this focus on computing infrastructure will determine the profitability and business model of chatbot generative AI. By not being subject to third-party chip providers, like NVIDIA, Google may have an edge over OpenAI in this arena.
Google’s Second Mover Disadvantage
Despite their positioning and computing advantages, Google does face some difficulties.
The most common hurdle for second-movers is that they’re playing a game of catch up in gaining market share.
Google’s in this category.
OpenAI’s Opening Line Up
OpenAI, as first mover, has already secured several companies as customers, including Snapchat, Microsoft, and Duolingo, among others to embed ChatGPT into their core products.
Each company is using ChatGPT in ways tailored around their business.
Microsoft has integrated ChatGPT into their search engine Bing. Snapchat embedded ChatGPT into a personal assistant in-app, “My AI”, and Duolingo is using ChatGPT to role-play different languages for users.
It’s unlikely that any of these companies would switch OpenAI’s language model for another anytime soon. After all, such a process would involve migrating their user data and redesigning their platforms to integrate with a new language model, not to mention training the new language model with the migrated data.
All of this requires colossal EMT (effort, money, and time).
Nothing in life is guaranteed
Even with a perfectly engineered transition, there’s no guarantee the language model switch helps the user experience.
Users, after becoming accustomed to ChatGPT, may find frustration in the new language model’s behavior and responses.
There’s also a chance the new language model only marginally outperforms the former in accuracy and articulateness.
A note on switching costs
These risks and drawbacks associated with switching providers are often referred to as switching costs. Switching costs explain why customers stick with inferior technology. Even if the new technology vastly outperforms, the cost of switching keeps customers glued to what they have.
The switching costs outlined above will likely make it difficult for Google to capture the market share ChatGPT has gained in its first months.
Unless Google significantly improves Bard or lowers their chatbot’s price, it is unlikely ChatGPT customers will flock to the tech giant for their chatbot services.
Conclusion
Though Google’s position as second-mover has allowed them to position their branding strategically, Google does face an uphill battle in contesting OpenAI’s ChatGPT.
ChatGPT’s growth and hype has established ChatGPT as a household name.
This branding will extend as more companies claim the rights to integrate ChatGPT in their product.
However, Google does have an ace up their sleeve.
The second they add Bard to the Google homepage, millions (if not billions) of users will flock to Bard.
Once Google does that, the whole game changes.
It’s only a matter of time.
Some Final Thoughts
If you’re reading this article, your world is likely going to be disrupted by AI.
We don’t know what the changes will be, but they’re certainly coming.
There’s a chance we all become human can-openers for robot overlords, collecting dust in a cupboard soon. There’s also a chance that AI unleashes ridiculous, ungodly levels of human talent, and we make the Renaissance look like an afternoon preschool session.
No one really knows.
At Byte-Sized Insights, we believe everyone should be informed of AI’s advancement, regardless if it’s good, bad, or sideways. Each week we post Medium articles on progress made in the technology and business of AI.
They’re simple, quick, and designed for you, the people.
Follow using the link below.
Click here to follow Byte-Sized Insights before the AI overlords make us can-openers