When five of the biggest tech companies form a partnership to crack artificial intelligence, it’s bound to send ripples of excitement as well as fear across the brave new world of AI startups. And that’s what we’re seeing – excitement over new opportunities this will open up, mixed with fear over innovations it may smother.
Yesterday in the US, Google, Microsoft, Facebook, IBM, and Amazon announced a Partnership on Artificial Intelligence to Benefit People and Society. These are five of the biggest players in this space, each vying to become a platform for AI by putting out open source libraries and frameworks for startups and innovators to build upon.
The new alliance aims to share AI research and give it a push to bring AI solutions for everyday problems into the mainstream. Each one of the big five is sitting on a stack of AI research, so shared learnings should give the industry a huge boost.
As the saying goes, the whole is greater than the sum of the parts. This has different manifestations here, one of which is interoperability between AI frameworks and stacks.
Benefits of standardization
This is illustrated by MindIQ, a Bangalore-based #startup that’s building up an AI engine for business chatbots. It enables any business to set up AI-powered chatbots within minutes for Facebook Messenger, Telegram, Slack, and even WhatsApp if it opens up to bots soon.
MindIQ has used multiple APIs (application programming interfaces) from Facebook, Google, and Microsoft, all of which have their own open platforms for AI. Facebook’s Wit.ai is better with interfaces, while IBM Watson is ahead in machine learning and natural language processing. Google has the edge with emotion analysis as evident in its newly launched smart messaging app Allo. So the MindIQ AI engine has been mixing and matching the best of what it requires from the three platforms, explains co-founder Anup Reghunathan, who earlier worked for 15 years at Microsoft, where he helped develop the intelligent personal assistant Cortana.
Patching the different AI stacks together is no mean task, because each has its own protocols. So Anup looks forward to new standardization from the alliance to make his work easier.
It’s akin to how VHS became the standard video format in an earlier era, points out Kaiesh Vohra, co-founder of Singapore and Bangalore-based startup Lucep that has an AI engine to generate smart sales leads and act on them quickly](https://www.techinasia.com/meet-artificially-intelligent-sales-gorilla-closes-deals-faster). He says standardization will also help large corporations acquire or collaborate with AI startups because their solutions will be easier to integrate.
But there’s a flip side. When five big AI systems come together, it will quickly create new possibilities which may suddenly crush innovations that startups are currently working on. For example, Google’s Optimize 360, which now has a free version, is going to impact several analytics startups in the customer relationship optimization space. The same thing could happen to AI startups with new free tools coming out from the big five alliance.
Kaiesh reserves judgement until more details emerge on what the alliance will do, but he can already see that it will pose some new risks for startups. For example, a startup building something on the alliance’s AI frameworks will have to publish its tech in order to be standards compliant. This gives companies like Google access to tech that they might otherwise have had to buy or acquire.
Apart from the tech, the only differentiator between AI engines is the data that goes into them. And when it comes to data, the likes of Google and Facebook have oodles more than any startup, except in super niche areas such as that of Lucep which is building a specific solution addressing the needs of small and medium businesses in sales.
“It will need a shift in thinking from startups, who will have to be even more agile and adaptive than before,” says Kaiesh. But even if it shifts the playing field for him professionally, the academic in him rejoices over the impact this could have on the field of AI as a whole.
Kaiesh, who studied AI at Edinburgh University, which has one of the world’s best AI labs, says the open AI frameworks of the alliance could help academics progress in their research and contribute more to commercial applications.
Anup of MindIQ hopes it will lead to better AI which in turn will help its spread in the mainstream. “Right now, you can’t have a real conversation with a bot because it becomes quickly apparent that it’s not a human and doesn’t respond like a human,” he points out. So if the alliance gives a push to natural language processing and other AI technologies, it could make a difference to its adoption.
New, new possibilities
New, new possibilities will emerge, and startups will be able to quickly build on robust open AI frameworks to come out with solutions. As the new CEO of Microsoft Accelerator in Bangalore, Bala Girisaballa, put it to Tech in Asia in an interview last week, “We are at the cusp of a new era where everything is about intelligence. It’s no longer about automation, it’s about tons and tons of data and what we do with it… Give it another ten years, I think artificial intelligence and data analytics will be in the DNA of every company.”
The partnership between Microsoft, Google, Amazon, Facebook, and IBM is a giant step in that direction. But there are two other major entities in AI which are conspicuous by their absence from the alliance – Apple which is developing its voice-responsive Siri, and Elon Musk’s OpenAI platform which has US$1 billion in funding.
Apple generally likes to go its own way. But Elon Musk keeping out of the alliance poses a separate problem, because he has been expressing concerns over the way AI is being deployed and the harm it could do.
Neural networks mimicking the brain for the self-learning AI engines behind speech recognition, computer vision, and natural language processing have been around in academic labs for decades. But it’s only now that AI is coming into the commercial mainstream, from consumer electronics to business intelligence.
The improvement in AI technologies along with open access to AI libraries have led to a spurt of startups in this space. Funding in AI startups crossed US$1 billion in the second quarter of this year, according to CB Insights data. One can expect a lot more activity going forward, as it becomes easier to build AI solutions.