3 Questions to Ask to Identify AI Impostors
August 13, 2019 / artificial intelligence, Business, Intelligent Process Automation
In the technology industry, it’s not unusual for vendors to want to latch on to the latest trend and claim to have a product or service that fits the category. Artificial intelligence is no exception, which means customers need to be vigilant about querying vendors to ensure their technology can really be classified as AI at all, if not intelligent process automation (IPA) specifically.
The London-based venture capital firm MMC found that of 2,830 startups in Europe that were classified as AI companies, only 1,580 – about 56% – actually offered AI technology.
“We looked at every company, their materials, their product, the website and product documents,” David Kelnar, head of research for MMC, told Forbes. “In 40% of cases we could find no mention of evidence of AI.” In such cases, he added, “companies that people assume and think are AI companies are probably not.”
The Forbes piece goes on to explain that the companies didn’t necessarily classify themselves as AI firms; they were classified as such by third-party analytics websites. But the companies weren’t quick to correct the classification either.
That makes the customer’s job all the more difficult as they search for intelligent automation technologies to solve real business problems. So, to do our part to help you separate the AI wheat from the chaff, following are three questions to ask prospective AI and IPA solution vendors to determine whether the technology they’re hawking is real.
1. What’s your algorithm strategy?
Some AI vendors have their own, homegrown algorithms that they’ve trained over time. Others use open source algorithms such as TensorFlow, that are open to the general public and are constantly being improved. Either way, the vendor should be able to explain what its algorithm does and where it came from. If not, run away.
2. What’s your data strategy?
Here the answers may vary. For Indico, the answer lies in our generalized model that is the baseline for all of our IPA automation tools. The Indico base model consists of more than 500 million data points, which is enough to enable it to understand human language and context. Users can then customize the model to take on whatever task they’re trying to tackle, but using 100x to 1000x less data than would normally be required. Here again, if a prospective vendor can’t articulate its own database story, perhaps they’re part of the 40% that aren’t really selling AI.
3. What’s your application strategy?
For AI to be useful, it has to come with some sort of application that makes it accessible to those who want to employ it, whether data scientists, IT, or business people. Indico, for example, has a point and click user interface that makes it simple for anyone – including business people – to build effective process automation solutions and models without the need for data science expertise. Without some sort of application like that to make the AI technology useful, you’re not going to get much value out of it.
AI startups are targeting process automation
As with most technology categories, there is a lot of noise, which makes it difficult to figure out which vendors can add real value and help you address the challenges you face. Hopefully these questions will help you find AI vendors that can solve real business problems.