The challenges with AI innovation
Most of the companies that I have been working with want to become a technology company somehow which is triggering a huge effort in terms of business and technical transformation capabilities.
Nowadays, the majority of CMOs (Chief Marketing Office) and senior management people praise and pitch that their companies are embracing and using AI across to improve their client service workflows in an inspirational way.
The challenge is that it goes beyond technology adoption. Human talent, culture and a holistic business view need to be in constant motion to shift directions quickly based on business hypothesis. For example, if your company is not dedicating a significant effort in understating its customer journeys and taking actions in managing customer services accordingly, you may expect bigger challenges in adopting or incorporating AI capabilities. Why? The challenges may vary from lack of business agility practices, experimentation culture and capabilities to reveals the client’s journey. It doesn’t matter if you are a technology service provider, start-up, ecosystem player or from traditional non-tech industries that now must deal with technology as a core engine in trying to stay relevant for their customers and innovate.
In such context, adding “AI” promises and benefits to your portfolio/product became a “must”, otherwise you may lose the momentum and pass the message that you may not be ready to explore current opportunities that will help you or your company to shape the future.
These phenomena may explain why Artificial Intelligence is one of the most misused terms in the industry today. According to the survey from London venture capital firm MMC, 40 percent of European startups that are classified as AI companies don’t use artificial intelligence in a way that is “material” to their businesses. The causes may vary from metadata classification from analytical websites to investors appeals to the theme that may drive contribute to the hype in the market now. Interestingly, this fact backs up Mr. Tripathy’s thought process.
In a recent podcast, Daniel Faggella (Dan runs one the best AI podcasts AI in Industry) has interviewed Abinashi Tripathy (founder and Chief Strategy Office at Help Shift — His primary focus is to apply AI to the future of customer services). Abisnashi has shared his view on the challenges of conversational interfaces such as chatbots, automated call services and so on. He has highlighted that it may take a couple of years until those AI capabilities are robust enough to a meaningful degree.
According to Mr. Tripathy, from a customer service perspective, AI is very good at doing classification but it’s not robust yet at maintaining a sustained conversation. Therefore, it is very effective in matching knowledge assets that a company has through a specific question and extract intents that can be applied to knowledge assets and curated content. He has explained that one of the problems with AI today is the core start problem which usually needs a ton of data to be tackled in case you apply deep learning techniques but considering recent advances in NLP, it is possible to extract intent off an incoming question in combining two statistical models and one sort of shallow learning which does not require huge amount of data upfront, specifically for text-based NLP.
One of the highlights from the interview is that both Google and Amazon have been investing and working on solving sort of the natural language challenges with an army of PHDs and seems they still have a ton of work to do. For example, Amazon with its Lex platform. They use Lex to power Alexa built-in as a consumer device that people can use in their homes with very limited use cases but they have not deployed this technology in their customer service chat yet whether it’s the phone channel or whether it is the chat experience because they know that the technology’s not ready to go beyond basic use cases.
According to Mr. Tripathy, technology will get to the point where the accuracy levels are very high and then we can shift from the sort of decision tree bots to a more human-like conversational bot. He foresees AI breakthroughs coming from big consumer companies like Google, Amazon, and Microsoft as they have the most extensive data and capabilities to validate models and algorithms that then can be used by startups in consuming those services as a platform instead of focusing on the core algorithms. He has also mentioned IBM work as an “on-premise” model due to its enterprise focus.
On the other hand, startups may focus on how to solve specific vertical problems or business cases as they do not own consumer data, human capital and resources do compute it.
In summary, large consumer data sets are the new gold of our century! It may also reflect the challenges to innovate with AI and the hype nowadays and also back up the interesting insights from the MMC report. Overall, we need a more realistic expectation about where AI can be applied and how to get the best of its usage. Even in other industries such as pharma, finance, telcos and retail that performs research and development may do prefer to join forces with big tech companies in connecting their data to these big consumer data sets and platforms to keep innovating.
I would love to know your opinion on this matter. Please feel free to add your comments.