Machine learning: 73% of enterprises are lost in survival

Everyone knows that machine learning (ML) is one of the key technologies of artificial intelligence and an application technology that is gradually becoming mature. Specifically, this technology can bring changes to future data science, allowing application companies to make driven decisions based on more data analysis, thereby improving users' business experience.

#So, in what aspects and to what extent has ML improved the business status of enterprises currently? Recently, Forrester Consulting based on a survey of 150 company data leaders and decision-makers in North America, and concluded some important performances of ML in business operation decisions. Which of these survey conclusions can help us and learn from us?

# Let’s first look at some key information.

• In the business affected by machine learning, automated anomaly detection (Anomaly Detection) is the primary task to be achieved in the next one to three years;
• On the technology implementation path, data silos, poor interpretability, and low transparency are the main obstacles hindering progress, thus slowing down the improvement of technology maturity. schedule.
• It would be more beneficial to focus more on business outcomes and to establish partnerships with companies that have a lot of practice and proven effectiveness in ML technology. implementation of this technology.

Only a quarter of ML applications are in the mature stage

In the development and development of machine learning In terms of release time, most respondents chose between 1 and 5 years, accounting for a total of 72%. Among them, more than half said their apps were released in 1 to 2 years. In fact, a mature machine learning strategy requires a precipitation period of three years or more. Only about a quarter of companies that meet this standard have applied it for more than five years, and only 5% of them have applied it.

Additionally, 53% of respondents plan to improve business efficiency by leveraging ML.

In terms of current big data and data analysis strategies, 46% of people choose to use multi-cloud (including private cloud); 44% choose to study stack performance , so as to better utilize data for model architecture; 41% chose to expand the scale to meet the needs of the increasing data volume.

#In the next one to three years, the main strategic application directions of ML are: automatic detection of abnormal data (40%), automatic transparent application Receiving and infrastructure updates (39%), as well as making AI applications comply with new regulatory and ethical requirements (39%), etc.

The most challenging thing to do is to solve data silos in technical management

Except Technical capabilities, machine learning also faces considerable challenges in personnel and process management. Among them, 41% believed that breaking down internal data silos was the most challenging, and 39% chose to convert academic models into deployable products. In addition, 38% chose to reduce AI risks and break down external data silos respectively, and 36% believed that the biggest difficulty lies in processing large-scale, diverse and chaotic data sets.

Whether it is data silos, model transformation, or data set chaos, they all reflect the gap between academia and commercialization. The gap, especially in the transformation of models, is that when using ML and extending it to use cases, many people find that the transparency, traceability, and explainability of the data flow are difficult to clearly present.

Because of this, when the prospect of ML implementation is unclear, management will believe that it is difficult to see business value in business implementation based on machine learning. And if there is no clear relationship with investment return, managers' intention to invest in this technology will significantly decline. 73% of the respondents believe that machine learning still faces challenges in data transparency, traceability and interpretability. The uncertainty of investment intentions has exacerbated the difficulties of technology implementation, and a virtuous cycle has yet to be formed.

Two-thirds of decision makers will still increase the application of ML

However, even in the face of many Challenges, decision-makers are inevitably wary when deciding to invest in machine learning, but most interviewees believe that the application of ML is still very necessary. Two-thirds of decision-makers (67%) believe that increasing the application of ML technology in an all-round way is very important for the organization’s strategic planning. 66% of respondents think it is important to add technical capabilities and applications of machine learning to the currently used toolset.

At the business level, the top three areas where machine learning is expected to play a role include: data platform mutual sharing, enterprise Track data flows within your organization and drive faster action.

As for cooperation with third parties, 37% of the respondents stated that they have established cooperation and intend to develop partners relationship; 30% said they have a cooperative relationship, but are not prepared to develop it into a deeper partnership. In addition, 19% and 11% of the respondents said they have cooperation plans or are interested in cooperation in the next year.

More than 60% of the respondents said that they use cooperative relationships to make up for their shortcomings in machine learning and personnel shortages, indicating that win-win cooperation is still It is an important way to develop this technology. Cooperating with third parties with experience in the field of machine learning can create synergy in model development, personnel training, and mining of more data sources.

Article reference and image source:

##Operationalizing Machine Learning Achieves Key Business Outcomes

The above is the detailed content of Machine learning: 73% of enterprises are lost in survival. For more information, please follow other related articles on the PHP Chinese website!