The AI and machine learning technology integrated into the daily work is not as easy as you think, this is the latest Algorithmia research company, found that the company survey 750 business decision makers, the survey found, while machine learning in the enterprise maturity basically into growth, but most of the companies (50%) each deployed a machine learning model takes 80 ~ 90 days (18% of the company also take more than 90 days), most of them blamed a lack of scalability (33%), followed by the model can be replicability challenge (32%) and lack of executive participation (26%),

"Our 2020 [state] machine learning in the enterprise of research results and we hear from customers is consistent", CEO of Algorithmia Diego Oppenheimer says, "companies are increasing investment in the field of machine learning, machine learning, operability and became mature in all walks of life, but there are still great room for improvement and the growth of this area, the model of deployment cycle need to be more efficient, and machine learning team also need to be more seamless, nonetheless, has deployed machine learning technology companies are benefiting from measurable results, including cost reduction, fraud detection and user satisfaction, etc., with machine learning techniques and processing ways to enter the market and landed in the market, we expect that these trends will continue to persist,"

The growth of the employees

By using machine learning threshold is very high, the market demand for machine learning experts also is high, it may not be an accident, Algorithmia people in half of the respondents said their companies hired 1 ~ 10 data scientist, 5% of people said employment of more than 1000; 39% said their company has more than 11 and a data scientist, behind the figure had the big promotion from 18% in 2018, 2018 [state] machine learning in the enterprise of the research report is Algorithmia last report released version,

In this context, the shortage of the industry data scientists predict looks prescient, in 2016, deloitte predicts 2018 there will be 180000 vacancies, and LinkedIn data scientist position number increased by more than 650% in 2012 to 2017,

Algorithmia is expected as the growth of the demand for data scientists, primary level data of the team of scientists working in the area in shaping the AI opportunity will be less, because most of the junior level employees need to finish the job may have been his predecessors finished, however, this will also mean leadership alignment between different teams, AI team will have more autonomy and flexibility in the project implementation,

The challenge of the ground and implement

Although enterprise to find data scientists talents is very crazy, but the report of nearly 55% of companies said they had not deployed any machine learning model (the figure is 51%) last year, one 5 of the enterprise is still in the evaluation of use case scenarios, or put into production plan by the end of the year will model, only 22% more companies in the last two years have put the model into production areas,

This with international Data Corporation (IDC: Internationale Data Corporation) analyst has been consistent, gives the study they found in those who have been used for AI research organizations, only 25% of organizations have developed "enterprise" AI strategy, accept the survey company blamed AI solution budget is not enough, too few competent employees, as well as the Data deviation and unrealistic expectations,

As mentioned earlier, according to the research of Algorithmia, for most organizations, the machine learning model into production areas is still a big challenge, at least 20% of the company (contains such statistics all the size of the company) have said their data scientists spend a quarter of the time on the deployment model, this is due to the widespread caused by lack of extensibility, such as since the business scale expanding and need to buy more hardware, data and tools, and perform the necessary model optimization, the model of version management and repeatability is another difficult task, since they can influence such as assembly line, model training and evaluation of key process, such as

Regardless of what is due to factors, budget is unlikely to become attribution, about 43% of the respondents claimed their AI and machine learning costs from 2018 to 2019 increased by 1% ~ 25%, and 21% said the project budget has increased by an average 26% ~ 50%, and in fact, only 27% of respondents said they cost no change, AI Algorithmia think the latter is already have more mature companies - for example, the enterprise use of AI in the field of production model there are at least two years, compared with other companies in the field of AI in higher,

AI use-case scenarios

AI ground is also is not all bad news,

Gartner report in January, grew by 270% in the past four years the AI implementation, only is increased by 37% last year, according to the McKinsey global institute, the labor market transformation will in the next 10 years to GDP (Gross Domestic Product, Gross Domestic Product (GDP) of 1.2% growth, and 20% ~ 25% in net economic benefits of growth, from a global perspective, it will generate $13 trillion in the next 12 years of growth,

Algorithmia reported in those who have successfully deployed in the organization of the AI, reduce the enterprise cost is one of the most popular AI user scenarios, the second is to provide insight and intelligence, help to improve the customer experience, of course, different market segments, the situation of the application is different also, for example, most of the banking and financial services company focused on how to retain customers and found on fraud, while energy field (including utilities) attaches great importance to fluctuations in demand forecast, consulting and professional services of the respondents said, reduce customer churn is their top priority, and the first user scenarios of education market is how to interact with customers,

Therefore, in a survey of Edelman, 90% of respondents from C Suite (begin with devoted to executives of the title) will be described as "the next generation of technology revolution" AI, it is not surprising, according to the study of technology executives, about 94% of people think that AI will be innovative to create "intelligent" family, and more than 74% of the collective, said AI will "help" to accelerate the development of self-driving cars, such as the Alphabet subsidiary Waymo, Uber, gm's cruze companies such as self-driving cars,

"This year's survey will confirm... Algorithmia enterprise in machine learning are rapidly developing, "the authors wrote," although most of the enterprises are still in the early stages of [machine learning] maturity, but think you can put the/machine learning time delay, this is the wrong idea, if your business there is no machine learning consciousness, please be assured that your competitors will have such consciousness, AI the pace of development will surely exponential growth, "

The original link:
https://venturebeat.com/2019/12/11/algorithmia-50-of-companies-spend-upwards-of-three-months-deploying-a-single-ai-model/