[ad_1]
"If your competitors are applying AI and find solutions to speed them up, they'll be doing it very quickly," says Deborah Leff, global leader and chief technology officer for Data Science and AI. at IBM, on stage at Transform 2019.
On their panel, "What the heck does it mean to" do the AI? "Leff and Chris Chapo, senior vice president of data and badytics at Gap, Inc., explored why Many companies are still hesitating or simply fail to implement AI strategies, even though the benefit inherent to larger companies has disappeared, and the paradigm has changed completely. AI, fast-paced companies outperform slow-moving companies of any size, and tiny, unnamed companies are now stealing market share from giants.
But if there is a universal understanding, that artificial intelligence empirically provides a competitive advantage, why only 13% of data science projects, one in ten, succeed- they actually to production?
"One of the biggest [reasons] Sometimes people think that all I have to do is spend money with a problem or put a technology in place, and success comes out of it, and it does not happen, " says Chapo. "And we do not do it because we do not have the support from the leaders to make sure we create the conditions for success."
Leff adds that data is also the other key player in the chain, it's a double-edged sword: that's what makes all these badyzes and capabilities possible, but most organizations are extremely quiet, with owners who simply do not collaborate and leaders who are not. facilitate communication.
"Data scientists have looked in the face and told me we could do it, but we could not access the data," Leff says. "And I say, does your direction allow this to continue?"
But the problem with data is that they always come in different formats, structured and unstructured, video files, text and images, kept in different places with different security and privacy requirements, which means that projects are slow at first, because the data has to be collected and cleaned up.
And the third problem, intimately linked to these silos, is the lack of collaboration. Data scientists have existed since the 1950s – and these were individuals sitting in a basement working behind a terminal. But now that this is a team sport and the importance of this work is now part of the fabric of society, it is essential that each member of the team can collaborate with everyone else: data engineers, data managers, employees. which include data science, or badysis, or BI specialists, up to DevOps and engineering.
"It's a big place that holds businesses back because they're not used to collaborating in this way," said Leff. "Because they take these ideas and turn them around, you're now asking an engineer to rewrite a data science model created by a scientist, so how does it work, usually?"
"Good," says Chapo, "it does not work."
For example, one of his company's first projects in the field of data science created size profiles, which could determine the size range and distribution needed to meet demand. Four years ago, the data science team entrusted the algorithm to an engineer. This one has been recoded in Java and implemented. Two weeks ago, they realized that it had been broken for three and a half years.
"It broke because no one owned it, we did not have the data science team to be able to continually browse the models, consider it an badet and have data operations to ensure it works Explains Chapo. "We are starting to bring these methods of work to life. But it's difficult because you can not do everything overnight. "
"One of the biggest opportunities for all of us today is determining how we educate leaders from across the company," said Leff. "Before, a leader did not need to necessarily know what the data scientist was doing. Now, the data specialist has come to the fore and it is actually very important for business leaders to understand these concepts. "
AI will not replace managers, she adds, but managers who use AI will replace those who do not.
We are beginning to see this awareness on the part of business leaders who want to understand how machine learning works, what artificial intelligence actually means for them, and how to use it effectively. And these leaders will be the most in demand, says Leff.
Another key to success, Chapo adds, is to keep things simple.
"Often, people imagine a world in which we realize this type of brilliant, whimsical, unicorn, sprinkled with goblins intelligence projects," he says. "The reality is, start simple. And you can really prove your path in complexity. This is where we started to show value more quickly, but also to help companies that do not really understand the data to feel comfortable. "
It's not necessarily the sophistication of the model at first, it's about creating a better experience for the customers. In fact, businesses are no longer competing with their closest competitors, but with the best customer experience offered by another, even in a totally different industry. If you can call a carpool service on an app in a matter of moments, you begin to wish the same level of experience when you call the bank, file an insurance claim or place an order online.
There are three ways to start and avoid becoming one of 87%, says Chapo. Choose a small project to start, he says – do not try to boil the ocean, but choose a difficult point to solve, where you can show demonstrable progress. Make sure you have the right cross-functional team to solve this problem. And third, use third parties and people like IBM and others to speed up your journey early.
Source link
