AI Development Council's third meeting turns to industry, with the government and Samruk-Kazyna told to pick pilot enterprises for AI transformation

Astana TV, citing the Akorda press service, reported on 2 October that the third meeting of the Council for the Development of Artificial Intelligence took place at the Akorda under President Tokayev, devoted to the deep digital transformation of industry.

The President opened with numbers: in the first eight months of this year industrial output came to about KZT 46 trillion, and manufacturing grew 8.4% on a year earlier. To keep that pace, he said, labour productivity has to rise, the industrial base has to be modernised and more high-value goods have to be made, and AI is a key tool for that.

He named several tasks. First, digitalisation at many enterprises is piecemeal: individual operations are automated while the production model itself stays the same, and that approach has run its course. The whole production chain needs rethinking to see where AI would add most, with real results as the main test. The government, together with Samruk-Kazyna, is to pick the enterprises whose key business processes will undergo full AI-based transformation, and the best models are to be scaled up nationwide.

Second, science must work for production. Qazaq AI Research University is to be not just a centre for training specialists but a leading site for applied research, with a programme tied to the economy's practical needs. That calls for a procedure to identify and keep updating industrial AI tasks, framed by people from production, engineers, technologists, foremen and operators. The government is to draw it up with national companies and private business.

Third, the AI Challenge mechanism is to draw in digital nomads and talented young people, with teams given access to data, computing power and production infrastructure, the customer involved at every stage, and results tested in real conditions. The President accepted that industrial AI and DeepTech projects take a lot of time and money early on, carry high technological risk and have no proven commercial viability yet, so the state should help cut initial risk and bring in private capital, after which promising projects should mostly develop on market terms.

Fourth, many solutions do well in pilots and then stall on extra approvals, repeat testing, cumbersome procurement and regulatory limits. The government is asked to design a fast track from pilots to commercial rollout, not to repeat the full testing cycle where conditions are similar, to give particular attention to domestic solutions that have proved effective, and to remove needless barriers where economic benefit has been shown.

Fifth, data. Enterprises have built up large volumes of production data, but it is scattered, uneven in quality and often unfit for AI. The government and business are to work out common approaches to sectoral industrial databases with uniform standards for quality, interoperability and safe use, clear access for domestic researchers, and protection for trade secrets, strategic information and critical data.

This is direction set at presidential level, and the piece gives no sums, deadlines or list of enterprises. Two points are worth Chinese companies' attention, "domestic solutions first" and "data openness within limits". They will shape how deep foreign technology suppliers can get into projects, but the actual rules are yet to be adopted.

Source: Astana TV (citing the Akorda press service), https://astanatv.kz/ru/news/179634/