In theory, the recent Google Next cloud computing conference was focused on cloud computing. As if to prove the point, Google pasted its marketing slogan all over the event in massive font: “the new way to cloud.”
In reality, I’ve never been to a cloud conference in which cloud played such a supporting role to another emerging technology – and I’ve attended numerous cloud shows.
So what was the other technology? You earn 5,000 bonus points if you can guess. Hint: high school seniors are using it to write their college entrance exams.
Okay, congrats – you guessed correctly. But given the tech trends this year it was no surprise: clearly, artificial intelligence was the true star of the Google cloud event.
This year some 8,000 attendees converged on the Moscone Center in San Francisco to talk shop, endure vendor pitches, and eat at some of the best (and most expensive) restaurants in the country. I’d be surprised if every last one of them didn’t at least mention AI as they mingled.
So what was everyone saying? To provide insight about where we are with artificial intelligence – and where we’re going – I spoke with six industry experts, including a range of vendors and analysts.
For analysis of what the Google Cloud conference means for cloud and AI, and to read the expert predictions, go to:
- The (Real) Big News: The Future of Cloud and AI
- Gopal Srinivasan, Principal, Deloitte
- Sheenam Gupta, Broadcom
- Jennifer Marchand, Capgemini
- Ted Kwartler, DataRobot
- Ed Anuff, DataStax
- Dean Pratt, Kyndryl
As the flurry settled, it was obvious: the biggest grand reveal was about Google’s enhanced partnership with Nvidia – in other words, the news that really counted was about AI.
The chipmaker Nvidia, of course, is at the very molten core of the AI frenzy gripping the tech world in 2023. Companies know they can’t obtain top AI performance without Nvidia’s powerful GPU processor chips. As the interest in AI has surged, Nvidia’s stock price has rocketed more than 200 percent this year; when Google announced the new alliance, Nvidia’s share price spiked 4.2 percent in one day.
The improved alliance means Google cloud customers can now more easily access Nvidia solutions. Google will host Nvidia’s DGX cloud, including the initial instances of DGX supercomputing, so Google customers can deploy supercomputing capabilities. Additionally, by next month, Google will offer A3 virtual machines powered by Nvidia H100 GPUs. In other words, if you need access to Nvidia’s limited supply of ultra-fast AI processor chips, Google Cloud is a source.
But here’s where things get blurry. Sure, the announcement was a big deal on the AI front, but it’s also major cloud news. The deal is a prime example of how cloud and AI are now joined at the hip.
The Google-Nvidia partnership reveals a major truth about AI’s future: AI can’t grow without cloud to support it. AI gets all the headlines these days, but in reality AI is part of a trilogy: artificial intelligence, cloud computing and data management all work together to support AI.
AI is a data hog; it requires vast repositories of data to do its work. And at the enterprise level, AI requires robust, scalable computing power – hence the need for a cloud platform like Google’s.
The fact that cloud shared the spotlight with AI at the Google Cloud show – an event ostensibly focused on cloud – is not a negative about cloud’s fortunes. Just the opposite: the event’s deep focus on the AI-cloud relationship demonstrates the primacy of cloud computing.
In the big picture, this cloud-AI symbiosis means that, in the battle to control AI’s future, the large cloud providers are largely in control. That’s good news for Google, Amazon Web Services and Microsoft Azure. It’s challenging news for anyone seeking to compete with these mega players – though there are no lack of game upstarts wiling to try, so we’ll see.
As for Google, the company was clearly happy to focus on AI at its annual cloud show. Among the big three in cloud, Google runs a perennial third behind AWS and Azure. If the battle for market share is about cloud alone, that works against Google. The company is smart to play up its traditional strengths in data and AI.
However, the extent of Google Cloud’s benefit from its Nvidia alliance is unclear. Cloud customers can also access Nvidia’s solutions through the AWS and Azure cloud platforms. And it’s probable that each cloud provider is nurturing its Nvidia partnership, so the Google-Nvidia offering might not be unique in the long term.
At any rate, the excitement in the tech world about Nvidia is frothy, to be sure. As I spoke with people at the Google conference, I heard numerous references to “Jensen” – the Nvidia CEO’s name is Jensen Huang. Amid the AI frenzy of 2023, he has become such a celebrity that people refer to him with one name, much like Drake or Beyonce.
Also see: 100+ Top AI Companies 2023
Vendors and Analysts: Expert Opinion
I spoke with the following experts at the Google Cloud event about the future of generative AI. Their comments have been edited for length and clarity.
Principal, Deloitte, Generative AI Leader for Google Alphabet Alliance
I’ll start at the high level and then get into specifics over how AI might evolve. We’ve spent the last hundred years trying to abstract away how we as humans interact with computers. And we’ve gone from vacuum tubes to software and the internet – and generative AI is a new leap in how we interact with computers.
It’s going to become all-pervasive and be in every form of computing, from software applications, software in physical devices, and within the enterprise, across the board in every function, from finance to HR to customer service. Generative AI is going to become the way we build applications and deliver experiences for users.
How might this play out over 1, 3, 5 years? That obviously remains to be seen, but there’s already significant momentum, and over the next 12 months we’ll definitely see a category of applications that are centered on essentially assisting users. Meaning to help perform a workflow: help them perform faster, get them the insights and the information they’re looking for faster.
If you go beyond that in the second phase, you’ll see that much more augmentation. And the difference in going from assistance to augmentation is that AI will perform portions of the workflow entirely by itself.
So humans will perform certain pieces, AI will perform certain pieces end-to-end, and the entire workflow will get strung together.
Beyond that is the third phase, autonomy, where entire work processes within the enterprise will be performed end-to-end by AI. For example, your entire finance reconciliation process, end-to-end, can be done by AI with no human intervention. So broadly speaking, AI is going to essentially seep into every part of how we use technology within the enterprise and bring to bear the next form of how we use computing.
eWeek: Do you think that today’s executives fully understand the technology enough to fully plan for it?
Not quite, but what I would say is, reflecting on the past couple of evolutions, there is definitely a lot more interest and willingness to learn. I think most executives get the potential, they obviously have a lot of questions about the risks, and they’re more keen on understanding how the technology actually works.
So we are not there fully in terms of executives getting to the full potential, how it can transform their business. But that’s always [true] with new technology paradigms, the process of education and change management. And that’s happening more rapidly than I’ve ever seen in my 25 years of working in and around technology.
Lead Product Management, AIOps and Observability, Broadcom
The one thing that I’m seeing settling down now, as compared to when we talked about AI and ML five years back, is that there’s a realization [among executives] that AI/ML is not magic. It can provide outcomes if it is aligned to the right use cases or the right goals for an enterprise.
And so ideally what AI and ML needs to do is to augment knowledge. To augment [staffers] skills to help them avoid doing the mundane day-to-day tasks and leave it to the insights from AI or ML. Instead, have the teams focused more on innovation or adopting the newer technologies because that will also take a lot of experimenting, just like AI and ML did.
eWeek: What about the executives who have some doubts that AI can really play a leading role?
Oh yeah, we hear that all the time from the executives, they say, “are you telling me that because you have AI/ML in your product, it’ll replace all the tribal knowledge that my teams have? Or will it help us do some cost reductions?”
And our answer typically is a yes and a no. A no, because what AI and ML should essentially do is really augment what their teams have been doing. It should really learn from them and should support them. Or it should provide that next level for them to do things faster or to make sure their customers are getting the kind of experience they want to deliver with the help of ML. And a yes, because it would help their existing teams to channelize or to expand their scope of influence across [areas] more oriented with their business goals instead of staying stuck.
So our answer to them is that our machine learning solution will be a supervised model, which can take feedback from their employees; it should partner with them rather than competing and replacing them.
Also see: Top Generative AI Apps and Tools
Google Cloud COE Leader, the Americas, Capgemini
One interesting area we’ve started to build out with generative AI for our customers is conversational commerce, so really about expanding the channel reach to customers with minimal investment.
Take for example, a kiosk in a store concept. So you’re able to put a digital-human order kiosk in a partner store, and it’s ‘always on’ for business. Our focus group found that consumers are more willing to engage for longer with technology when you have the digital-human front end. And when you layer on the generative AI component to that, it creates much more dynamism with the consumer as well as the ability to upsell, cross-sell, and personalize that experience.
eWeek: What about generative AI’s effect on jobs and employment?
I think that jobs will evolve and the people who will succeed are the people that embrace and learn how to use generative AI to make themselves more productive and competitive. But it’s all about finding and designing the use cases to keep the human in the loop.
Figure out the things that you don’t want to do manually. Give that to generative AI and add your value where it’s really going to have an impact.
Field CTO, DataRobot
I think in the next one to three years, you’re going to see a merger of predictive AI – that “good old-fashioned AI” – with generative AI.
I think there’s going to be a merger of the two because we’ve seen that consumers want to interact with machine learning and AI in a way that’s easily consumed and contextualized for them. Bringing the two together makes a lot of sense.
In the interim, where we’re at right now, people have to make the choice of what model to use, and if they go off in the wrong direction, given the rate of change and the rate of innovation, they may end up building in a lot of tech debt.
So I think in the 12, 24 month timeframe, people that succeed will have to have a lot of flexibility so that they can avoid tech debt. I think in the end, we end up in a place that’s very valuable to the enterprise. I think we merge predictive and generative so that the end user has more context for what’s getting forecast.
Also see: Generative AI Companies: Top 12 Leaders
Chief Product Officer, DataStax
When we’re talking about the enterprise, it’s really interesting because the truism of the future being unevenly distributed is going to be valid in this case. We’ve got folks that are going to be doing all the stuff that people are talking about: chatbots and AI assistance live on their websites this year. But the majority of folks will be getting this stuff into production next year and the year after.
So the “crossing the chasm” still applies. However, if you’re not doing everything that everyone seems to be talking about, don’t worry about it. There’s still plenty of time to build the right things for your business and get them out there.
I do think costs are going to keep going down, and the power of the computation, whether it’s GPU or CPU-based, is going to keep going up.
The options for running these [large language] models are going to get more varied. You can now run Llama 2 on your laptop and people are doing it and getting great results. And if I can run Llama 2 on my laptop, it means I can run it in my data center as well. So we’ll see a lot of that.
We’ll see just massive new things in the next wave of these models. The stuff people are talking about in terms of their size and power – a lot of it is unpredictable. A lot of the stuff that we’re dealing with now is just going to get a lot more powerful and easier to apply.
But more power, lower cost is going to be the main thing. The tooling is going to get better. More of your developers are going to be able to use AI – not just your data scientists or your ML experts.
eWeek: What about the high cost of hiring AI experts?
Look, most of us who have been around for a while, we remember 10 years ago, 15 years ago, when every major brand, every enterprise, everyone in the Fortune 500 needed to build mobile apps and mobile developers were very expensive at that point. And many companies farmed out their mobile app development to consultants, digital agencies, and paid top dollar to do it because the opportunity was that pressing.
So I think the same thing is happening with AI right now. There’s inarguably a lot of skills that you need to make this happen: top-notch experts. But again, that’s a point in time. The tooling is getting better, so then the costs will go down.
A lot of people are surprised by that because there’s a lot of mystique around data science right now, and it’s improving very, very fast. If you look at a lot of these startups that are building these amazing apps on top of AI, these are people who are in their first exposure to AI. They had no previous experience prior to this year, yet you’re seeing them build these amazing things.
Google Cloud and Generative AI Evangelist, Kyndryl
Where I see generative AI headed, I think it will accomplish some of the technical skills that are needed today in a much faster way. We already have NLP to [create] code. We already have generative AI accomplishing a lot of the functions and summarizations and data gathering that we look for.
However, with all the new possibilities, there’s a wealth of pitfalls as well. New tools are extremely important, but understanding those in a responsible manner – being aware of the risk exposure when utilizing new tools as well as the beneficial outcomes – it’s important.
Where I’m excited with generative AI is in the diametric conversions of quantum computing and generative AI, and the expansive neural matrices we’ll see as a result – this should provide some very exciting outcomes.
We’re going to see the industry shift in monumental ways – I’m very excited to see that and also help lay the foundation for the responsible adoption of those tools.
Read next: Generative AI Startups