How is Amazon — the company that plunked the Alexa AI device into more than 500 million homes and offices — responding to the tsunami of generative AI offerings touched off by ChatGPT?
In recent months, the company launched the Q chatbot for businesses and Rufus, a shopping-focused chatbot. But there’s a sense that Amazon is racing to keep up with Microsoft, its neighbor in Seattle, as well as Google and OpenAI.
“Amazon was hampered by the fact that in their heart of things, they’re an e-commerce company, and there is a principle that when you’re pitching new projects…there needs to be one portion of your pitch which addresses how it will spin other Amazon flywheels,” says Jay Baik, the former Head of Amazon Alexa New Business Incubation at Lab126. Lab126 is the company’s secretive Silicon Valley incubator that gave birth to Alexa.
Baik previously worked in advanced product design at Juul Labs and in product innovation at Samsung. We spoke about Amazon’s culture, AI, and corporate innovation broadly as part of a new research initiative, “How AI is Influencing Corporate Innovation Priorities.”
• • •
OpenAI’s Release ‘Surprised Everyone’ — Including Amazon
When I say Alexa and I talk about Amazon, I’m both using them literally but I’m also using Alexa as a shorthand for the state of all AI agents at that point [of] the introduction of ChatGPT [in November 2022.] It’s not like Siri was a lot better than Alexa or Hey Google at that point. It [was] all sort of in a similar state of capability.
But I think to be fair and to throw everyone a bone, even scientists today: they don’t know how ChatGPT does what it does. It continues to surprise people with new things that they just didn’t know that it would do. And the thing that stymies everyone is, at the end of the day it simply predicts the next word, based on all the data that it has been trained on. So I think in Alexa’s defense, it’s like no one knew unless you’re OpenAI or a couple of other companies. It surprised everyone.
And there is a little bit of hubris — hubris melded with defending what they have. …Amazon culture is a little bit to blame. You may know that Amazon is very religious about running the company based on their sixteen leadership principles. All the reviews are based on it… In meetings, they are really invoked. I’ve never looked at a company where leadership principles are sort of like this canon…
They also have their business mechanisms, which again are revered equally. And so it doesn’t leave a lot of room…I think it’s great and it’s fine when your original DNA is an e-commerce company where there are knowns and there are things you can rapidly test and adjust — like A/B testing. Does this copy work better? Where should we place this button? And we can A/B test that quickly… Fine.
But you take those same institutions and now you apply it to an innovation space — like [the] Devices and Services [group] — and the same values and functions that were optimized for e-commerce may not be the optimum set of mechanisms and principles to serve a different kind of organization.
But it’s Amazon. It’s a little bit cultish in the sense that it is the cult of Bezos and what he did. They stuck to it, and it led to a certain rigidity of process and thinking that didn’t let them really consider building something or exploring something like OpenAI.
‘We’ll Deal with the Transformation as it Happens’
…When I joined [Amazon] in July of 2021, they were looking at innovation and then growth. …I don’t even know if transformation was on their radar, but when OpenAI came out, transformation definitely became paramount: “This makes us totally rethink how we need to do our business,” right? And they were just starting that journey while I was there because I don’t even think they realized this totally removes the need for certain teams. It was almost like “we can’t stop to think this through. We need to get to market with something first, and then we’ll deal with the transformation as it happens.” In my time at Amazon [around] the release of ChatGPT, it was not going to be an initiative-taking transformation, but a reactive transformation.
…Alexa was so cumbersome both to develop, deploy, and use. [So internal teams at Amazon were] racing to do what ChatGPT can do, and in doing so, there will be entire teams…that are no longer necessary because of the emergence of ChatGPT. One problem with Alexa, as you may be aware, [is that] you have to be fairly precise in what the command is, because if it falls outside of Alexa’s vocabulary, Alexa won’t know what to do.
…There are going to be huge repercussions not just for Amazon and Amazon Alexa, but any organization that has to transform around a generative AI-centric offering instead of a [Version] 1.0 voice agent offering.
But ChatGPT can now cut through that: I can be imprecise, I can use slang, I can use whatever I want. ChatGPT will understand what you’re saying and distill it down to a command. There are entire scores of engineers working on natural language recognition and other approaches. With ChatGPT, they are potentially no longer needed… So there are going to be huge repercussions not just for Amazon and Amazon Alexa, but any organization that has to transform around a generative AI-centric offering instead of a [Version] 1.0 voice agent offering.
Why Microsoft Partnered with OpenAI
[Microsoft] had the right leadership. Amazon is not known for their innovation. I forgot where I read it, but [someone wrote that] no tech company can survive two paradigm shifts in their own space. You think about Microsoft, and they missed how search would transform their space. And Google brought it upon them. They almost totally missed the cloud, but they’ve managed to battle back. And I would think having been burned twice, now they’re sort of thinking, “We can’t be burned a third time. So we need to be a little bit more future thinking and open-minded and start investing in different things.”
Amazon was hampered by the fact that in their heart of things, they’re an e-commerce company, and there is a principle that when you’re pitching new projects…there needs to be one portion of your pitch which addresses how it will spin other Amazon flywheels. And so the thinking is “Will my business help other Amazon businesses”?
…If you can connect this function, this feature, or this initiative to increasing Amazon Prime subscriptions, or increasing Amazon Prime sales…then your proposal or your idea will get more attention and more lift.
And so for Alexa, if you can connect this function, this feature, or this initiative to increasing Amazon Prime subscriptions, or increasing Amazon Prime sales, or boosting sales for Amazon businesses, then your proposal or your idea will get more attention and more lift. It will face less headwinds. This is such an ingrained part of the mentality, especially within Alexa, because Alexa [is] losing money hand over fist. If you have those blinders on, then you’re sort of self-selecting the kind of projects that people will bring to the table.
…If you are approaching something like an OpenAI/ChatGPT kind of construct, you are just facing so many hurdles, right? And the only people that would really have the wherewithal to do something like that are the actual AI R&D team. …It could only come from someone who’s really, really deep in AI tech. But the way they’re being used is to just improve Alexa under the current paradigm and framework of how Alexa works. And so unless they have the latitude to explore something like ChatGPT — unless it is a top-down mandate from top leadership, like “I want this, make it happen, and make it better” — it is really hard for a group to do that bottom up.
The Future of AI: ‘This is Going to Wreak Havoc’
…This is fundamentally different, and it fundamentally scares me. Prior hype cycles [focused on things like] IoT, smart home, “quantified self,” or AR/VR. This is fundamentally different, and it is going to displace people from existing jobs. It’s going to redefine current roles; it’s going to eliminate certain roles. I feel like this is going to wreak havoc. And I heard this one interesting podcast where they commented that one reason AI scares us so much — and it should — is [that] if AI were invented in some Scandinavian country that had universal basic income and a healthy social welfare net, I don’t think AI would scare us as much. But because it was invented in Silicon Valley, [in] the most brutal capitalistic country in the world, we have a different lens on how to view AI.
The Boom and Bust Cycle of Innovation Teams
Innovation functions and innovation organizations within larger companies…always go through a boom-and-bust cycle depending on the economy. It is no surprise that when the economy takes a downturn, innovation functions are the first ones cut.
…There is an increasing mandate to turn a profit. It is fine to do sky-high innovation or future thinking, but when the rubber meets the road, it is about how we can turn a profit.
So with that in mind, there is an increasing mandate to turn a profit. It is fine to do sky-high innovation or future thinking, but when the rubber meets the road, it is about how we can turn a profit. And so there is a shift from wide open innovation to actionable, monetizable innovation projects and that sort of direction.
Advice for Other Innovators
The saying, “success has many fathers, but failure is an orphan” definitely applies to innovation at large companies and can be pro-actively applied. Innovation at large companies means challenging the status quo — whether it’s in manufacturing, sales, marketing, etcetera, where the devil you know is often preferred to the devil you don’t. So it’s critical you win allies, address concerns and create a sense of co-ownership with internal stakeholders to remove as much internal friction as possible. Many innovators often think an innovative idea on its own merits will drive support and consensus but this is rarely the case, esp at big companies. It’s critical to find allies (i.e, “fathers”) to ensure your innovation efforts see the light of day.
