AI wars: giants battle as generative AI use soars
The exploding interest in generative AI tools like ChatGPT has forced companies across sectors to consider the impact of AI on their business models — both as a time saver and as a potential disruptor. But while opportunity abounds, many companies remain wary.
It took just two months from its November launch for the generative AI tool ChatGPT to hit 100 million users, making it one of the fastest growing digital applications of all time.
In a few short weeks the tool, backed by OpenAI, made global headlines, secured billions in funding from Microsoft, and saw soaring interest for its API from SaaS providers seeking to integrate the large learning model behind ChatGPT into their services.
But that was just the beginning.
Major generative AI competitors have responded with their own early-stage AI rivals, including Bard, from Google, Claude, from Anthropic, a team that includes former OpenAI researchers, and a text-to-video AI from the team behind Stable Diffusion known as Runway.
Investors are also excited.
But AI is an overnight success story that has been years in the making, says Baker Tilly Canada Digital Technology and Risk Advisory Partner Deepak Upadhyaya, who has seen clients finally click to the value of technologies developed over the past decade.
“I think what we’re starting to see is that a lot of people had heard about AI, but couldn’t conceptualise it and couldn’t apply it to their business,” he says.
“Suddenly they can see the value.
The ability to take advantage of this very powerful technology will be transformative for the mid-market. – Deepak Upadhyaya
“There’s so much investment going on in all kinds of AI, not just in ChatGPT and natural language processing, but also in computer vision and generative AI.
“I think the ability to take advantage of this very powerful technology will be transformative for the mid-market, but it means helping our clients work through what they can do to gain a strategic advantage for themselves.”
Anestis Dimopoulos, Head of Regional Digital and Risk Advisory Services at Baker Tilly Greece, agrees.
“It’s like the consumerisation of Web 3.0,” he says.
“Clients can actually see this happening and see what this means in terms of using AI.”
The clear opportunity presented by the field of new tools is remarkable, Mr Dimopoulos says — even if few clients have progressed to implementing AI at this stage.
“If you think about industries like the medical sector, for example, it’s something that can work hand-in-hand with a doctor, and faced with data and the symptoms of a patient, AI can produce a first evaluation,” he says.
“For banks, investment firms and insurance companies, they can be using AI to help assess a client from the credit risk side, or in monitoring for non-performing loans.
“For manufacturing, or products and retail, when they have a new concept, and have a new design, they can produce and evaluate it, reducing time to market. That will be useful as well.
“There are implementations for every industry, from media to architecture to anywhere. I think everyone’s exploring how it can be used right now.”
Caution tempers AI enthusiasm
Jan-Willem van Essen, a senior manager in IT Advisory services at Baker Tilly Netherlands has also seen widespread experimentation with generative AI among his client base, but says most businesses are still playing with the technology — and few have moved to formalise its use.
“There’s a movement underway where companies want to be sure that the AI algorithm is doing the right thing, so for ChatGPT there remains a bit of reluctance,” he says, warning that key questions for business are not yet answered.
“If you want to use it to create insights for your business, you will also want to know how those insights are created, what’s behind them, which data is being used.
“There’s also another side of things where people want to understand what happens to their data.
“Can Google or OpenAI leverage my data to deliver insights for my competitors?
“If we have software companies embedding this kind of technology in their ERP systems, how do we know that our data can’t be used to deliver our information and insights outside our business?”
Effective use requires good governance
Mr Upadhyaya says clarity around the input of corporate and commercial data, the way it will be processed and repurposed, and how the eventual output will be used are all key decisions businesses have to consider.
On the one hand, generic data that lack specificity to the company in question will be less useful than if detailed corporate content is used to train the models.
On the other, sharing commercially sensitive data with non-proprietary AI models poses some risks.
“One of the big steps for a company is to understand the specific use case — so for the case of ChatGPT, how can you use the AI engine to ensure it is your knowledge and information about your products or business or services that guides the response,” Mr Upadhyaya says.
“How do you make sure it is your data that is ingested and your information that appears in the curated answer?
“If you restrict the data in-flow you are limiting the capability of AI, which should be able to disseminate content from a large number of sources.
“But if you don’t review or limit the content, how do you ensure that the results that are coming out are appropriate for your location, your context and your culture?”
The answer, he believes, lies in the corporate AI strategy.
Companies want to be sure that the AI algorithm is doing the right thing, so for ChatGPT there remains a bit of reluctance. – Jan-Willem van Essen
“What is happening now — with all these big companies putting so many resources into AI and embedding it in products — is that you are going to start seeing and using embedded AI regardless of whether you have a well-defined AI strategy or plan,” he says.
“Effectively you are going to be forced to enter the marketplace because it is going to be in your search engine, in your Excel, in your Word and Outlook.
“It will be in the next iteration of all SaaS products, in your accounting software like Xero or Quickbooks, undertaking modelling behind the scenes.
“Whether you are ready or not, it is coming down the pipe.”
Baker Tilly Canada’s advice to clients is to start now to define a strategy and get ahead of the wave.
“Think about your governance for data,” Mr Upadhyaya says.
“Think about the protocols for its use.
“How will you govern the use and training of AI? What will be the rules and procedures behind it?
“You don’t want to be on the tail end of suddenly finding these products all have this capability, but you haven’t understood how to utilise this capability to create a strategic advantage for yourself.”
The clash of the giants
While generative AI technology could disrupt a vast number of roles and business models, it also threatens to upend the established dominance of digital giants, including Google.
Microsoft Bing, never a strong competitor to the world’s most powerful search engine, is now seen as a serious competitor in the race to incorporate AI effectively in search, after launching its ChatGPT-enhanced service in limited Beta in February.
It reportedly issued a ‘code red’ to concentrate corporate attention on the threat posed by ChatGPT late last year, and to accelerate launch of its own AI products.
And yet when Bard, its AI-enabled search engine was shown to the public in early February, shares in Google parent Alphabet plunged over a perceived error in the Bard demo.
What would a world with a less dominant Google look like?
It’s an expensive question to ask.
An estimated $73 billion is spent annually just by US companies to optimise their position in Google searches, or to ensure their content is surfaced when someone poses a question online.
That’s before the nearly $225 billion spent last year directly with Google on AdWords, to secure display ads and product listings at the top of the page.
Changes to the landscape for online search — which has remained stable for more than 20 years — could have unexpected consequences for businesses.
It’s estimated organic and paid search are responsible for 80 per cent of traffic to websites. Already websites that fall out of early search results can see declining online traffic.
Embedding AI tools in search engines might exacerbate this challenge, with company websites effectively vanishing from view as AI-enabled search engines provide answers and increase the incidence of so-called zero-click search.
Take the example provided by Microsoft of searching for a pet hair vacuum cleaner using AI-augmented Bing.
While information comparing vacuums might once have been provided by online stores to gain a competitive advantage in search, that information can now be scraped and surfaced by the AI — and shown next to advertising from competitors.
“The question becomes, what do I need to do to be on top in the search engine because that’s changing minute by minute,” Mr van Essen says.
“I might be on top for a couple of seconds or I might fall back to page three, four or five — and no one will ever find me.
There are implementations for every industry, from media to architecture to anywhere. – Anestis Dimopoulos
“For online companies, it’s really a struggle to understand how the rules are changing and what they will need to do.”
The stakes are high in the race to rewrite the search experience.
Despite the limited use currently of Bing for search, Microsoft has told investors every percentage point gain in market share can translate to $2 billion in additional advertising revenue.
Mr van Essen says it is not yet clear who will come out on top.
“At the moment everything is changing too fast. Microsoft has launched this. Google has done that. But the impact on the companies will be very different,” he says.
“If Microsoft loses this battle for search, then, okay, it was a nice adventure. But if Google loses this battle? Their whole business is gone.”