23 March 14:19
“By turning off AI, a business can fall to the sidelines of corporate history,” this forecast for the development of innovative tools was voiced by Sergei Golitsyn, head of T1 AI (IT holding T1), at the Alfa Summit.
Among the leading experts in the development and implementation of neural network technologies, he took part in the discussion “Artificial intelligence and big data: how to quickly move to results in numbers.”
According to Sergei Golitsyn, the main customer request when implementing AI is related to cost reduction. And the proven economic efficiency of AI solutions is already measured in hundreds of millions of rubles. He gave an example from the mining industry: using computer vision, the IT system identifies potentially dangerous objects that can paralyze the operation of crushers.
“The costs of launching such projects are relatively small: setting up an AI model, installing cameras – but significant savings are achieved during implementation,” the expert said.
The greatest effect on a corporate scale, according to Sergei Golitsyn, comes from the implementation of platform and MLOps solutions. In the banking industry, in his opinion, this is already a standard, without which businesses simply cannot withstand competition.
However, this does not happen with every project: some pilots have to be stopped at the start. Golitsyn called on business not to perceive every failed project as a disaster.
“If the pilot was not considered successful, and the team did not go into the project, realizing all the risks, this saves from wasted investments and growing distrust in the technology,” he emphasized.
According to the expert, the key reason for many failures is a weak digital culture, including the absence or insufficient use of analytical services. This often leads to the failure to formulate the target result for the AI pilot and a change in the requirements for the future tool during the implementation of the project.
Another reason for unsuccessful AI implementations may be the team’s lack of competencies. According to Sergei Golitsyn, the task of technology leaders is not to stifle the initiative of employees experimenting with AI tools, but to build safe MLOps pipelines inside and outside the company, allowing specialists to legally implement projects in new areas. This becomes easier as access to complex AI models is democratized as much as possible.
“Usually, innovative technologies first became available to the largest corporations, and only then moved down. But now the opposite is true: it is in many ways easier for individuals to gain access to modern models than for big businesses,” the expert said.
Related links:
