Experts from BCG and Gartner said that the test scale of artificial intelligence has been very large, but it is still challenging to obtain value from deployment. They suggested it and business collaboration as a solution.
There is no doubt that artificial intelligence has brought great potential to global enterprises. However, defective strategies, poor process change methods, lack of expertise and a general lack of understanding of technology still make many enterprises unable to obtain value from artificial intelligence.
According to a survey of 2500 enterprise executives conducted by MIT Sloan Management Review and Boston Consulting Group (BCG), less than two fifths of the companies that invest in artificial intelligence say that they have made corresponding business returns. Artificial intelligence includes machine learning (ML) and natural language processing (NLP) and other related technologies, both of which aim to imitate human thinking.
As reported by the media, there are indeed technical barriers in the process of implementing artificial intelligence. Therefore, experts from BCG and Gartner discussed five pitfalls and successful solutions for deploying artificial intelligence in enterprises.
1. It led artificial intelligence leads to a waste of opportunities
Shervin khodabandeh, partner of BCG in charge of gamma AI business, said that many enterprises are letting it guide the development and deployment of AI. The way they treat it is very similar to the way they treat ERP systems. Khodabandeh said that this was a major mistake because general AI solutions did not help the business.
When the IT Department of a customer of BCG spent $85million to purchase the top-level ml stack and modern architecture, the investment produced only minor improvements to the website and applications, khodabandeh said for example.
Instead, organizations should align their AI initiatives with business strategies. This means that the AI strategy should be led by the CEO or the business unit that intends to benefit from the technology, not by the IT department. Doing so will help identify where AI can increase competitive advantage, and ensure that the correct process changes are made so that AI production and consumption are consistent. According to the data of BCG and MIT, 88% of the respondents said that AI had an impact on their business, and they combined their AI plan with digital strategy.
2. “technology trap”
AI led by IT department often has a narrow technical lens, and the resulting AI is a “black box” solution, which can hardly provide visibility on how the system gets its recommendations. Therefore, the enterprise avoids the solution because it does not understand “what the system can do for me”. The MIT and BCG report shows that only 17% of the companies with CIO in charge of AI see value, while 34% of the companies directly put AI under the CEO.
The report recognizes from those companies that gain value from AI that AI should not only be a technical opportunity, but also a strategic initiative, which requires investment in AI talent, data and process change. The AI strategy should be viewed as a whole, rather than just seeking to explore what this technology can do.
You may have seen movies before: it was introduced into shiny technical objects and construction proof of concept (POC), but these proof of concept did not gain attraction and generate business value. In addition, organizations undergoing business transformation have limited bandwidth for such experiments.
“Trying to build some AI ideas in POC and see if it works — we call it ‘POC syndrome’ — is a waste of time,” khodabandeh said. “The business unit often says that we have tried, but never expanded.” Similarly, this is a serious failure to tie technology and enterprise achievements together.
Instead, you need to cut back on experimentation and start small. Successful enterprises can always focus on a few key business priorities, and AI can also help promote growth, improve profit margins and create competitive advantages. These efforts should be consistent with business transformation. “If successful, these ideas will have an impact in a meaningful way, and the organization will act around these ideas,” khodabandeh said.
4. the gap between talents and knowledge hinders the adoption of artificial intelligence
Due to the talent gap, POC is often unable to obtain enough attraction. Gartner analyst Tracy Tsai said that this may mean the lack of technical talents who can cooperate with AI and understand its value, or the lack of AI product managers who can not convey the unique product value. Sometimes, even data scientists cannot fully simulate how artificial intelligence will boost business.
Access to AI tools also varies. In some cases, lob (line of business) may have an internal data scientist who can build Ai solutions with the support of it. But sometimes lob and it do not have a team of data scientists, so they use AI solution providers. In these cases, lob usually makes its request, but it may be difficult to clearly express its requirements, so it needs to rely on it to select solution partners and develop business use cases. In some companies, AI projects are driven by it, and POC is constantly added to the lob cycle.
No matter which way the company goes, Tsai said that it, lob and data scientists all have the responsibility to agree on expectations before establishing POC. This means that the ontology and classification of the extracted data need to be consistent; Including how to interpret the input; And the output of ML model. To do this, companies need to invest in talent by recruiting, upgrading skills, and re hiring AI producers and their consumers.
5. failure of process change
When implementing AI, some companies tend to ignore the required process changes. For example, when enterprises use artificial intelligence to automatically collect consumer data for promotional activities, they may be usurping some marketing functions. The team may reposition and focus on creating new consumer experiences. But not all companies have prepared their teams for such subversion.
Ultimately, “companies that derive value from AI view process change as the core pillar of their business strategy and closely integrate their AI strategy into their overall business strategy,” kodabandeh said.
Process change involves aligning the production and consumption of AI, which requires strong cooperation between business, process, strategy, data science and technology teams to create AI suitable for specific purposes, khodabandeh said. It is helpful to create a cross functional team in the center of excellence dedicated to guiding such a process.
“Artificial intelligence is an important strategic opportunity, but it will also be a major strategic risk if the company does not consider it carefully,” khodabandeh said. “Enterprises must seriously integrate AI into their core business strategies and business processes.”
Responsible editor: CT