https://sputnikglobe.com/20260603/researcher-explains-why-ai-shouldnt-be-blindly-trusted-1124249869.html
Researcher Explains Why AI Shouldn't Be Blindly Trusted
Researcher Explains Why AI Shouldn't Be Blindly Trusted
Sputnik International
People often make irrational decisions, and while artificial intelligence can help predict and overcome these mistakes, it should not be trusted blindly... 03.06.2026, Sputnik International
2026-06-03T15:52+0000
2026-06-03T15:52+0000
2026-06-03T15:52+0000
beyond politics
russia
ai
artificial intelligence (ai)
st. petersburg international economic forum (spief)
https://cdn1.img.sputnikglobe.com/img/07e7/05/08/1110205255_3:0:1454:816_1920x0_80_0_0_1fcf5a9f4972c2af13ac1979e47fbfd9.png
According to Suvorov, one of the most common cognitive biases is overconfidence.Another vulnerability is confirmation bias — the tendency to accept information that aligns with existing beliefs while rejecting evidence that challenges them."We readily believe information that fits our worldview and are often hostile to information that contradicts it. This bias makes it harder for us to change, learn and absorb new information," Suvorov explained.A third weakness is the tendency to undervalue the future."We all live in the present, so it's difficult to appreciate how today's decisions — such as failing to save money — can make life harder down the road," he said.These patterns are the subject of behavioral economics, which studies predictable deviations from rational decision-making.According to Suvorov, AI can help researchers identify such biases and develop tools to mitigate them. However, AI systems have shortcomings of their own: they can hallucinate, make mistakes and are often designed to please users.Suvorov warned that as AI becomes more accurate, people may become overly reliant on it."It makes fewer mistakes than before, but that creates a more fundamental risk — the temptation to hand over increasingly complex tasks and stop paying attention ourselves," he said.The St. Petersburg International Economic Forum is taking place from June 3 to 6.
https://sputnikglobe.com/20250814/whos-the-author-scientists-assess-role-of-ai-in-text-creation-1122610086.html
russia
Sputnik International
feedback@sputniknews.com
+74956456601
MIA „Rossiya Segodnya“
2026
Sputnik International
feedback@sputniknews.com
+74956456601
MIA „Rossiya Segodnya“
News
en_EN
Sputnik International
feedback@sputniknews.com
+74956456601
MIA „Rossiya Segodnya“
Sputnik International
feedback@sputniknews.com
+74956456601
MIA „Rossiya Segodnya“
russia, ai, artificial intelligence (ai), st. petersburg international economic forum (spief)
russia, ai, artificial intelligence (ai), st. petersburg international economic forum (spief)
Researcher Explains Why AI Shouldn't Be Blindly Trusted
People often make irrational decisions, and while artificial intelligence can help predict and overcome these mistakes, it should not be trusted blindly because it has vulnerabilities of its own, Anton Suvorov, rector of the Russian Economic School and a specialist in behavioral economics, told Sputnik at SPIEF.
According to Suvorov, one of the most common cognitive biases is overconfidence.
"For example, 80% of drivers believe they drive better than average. Among professors, the situation is even more striking — 95% think they lecture better than their colleagues," he said.
Another vulnerability is confirmation bias — the tendency to accept information that aligns with existing beliefs while rejecting evidence that challenges them.
"We readily believe information that fits our worldview and are often hostile to information that contradicts it. This bias makes it harder for us to change, learn and absorb new information," Suvorov explained.
A third weakness is the tendency to undervalue the future.
"We all live in the present, so it's difficult to appreciate how today's decisions — such as failing to save money — can make life harder down the road," he said.
14 August 2025, 12:05 GMT
These patterns are the subject of behavioral economics, which studies predictable deviations from rational decision-making.
According to Suvorov, AI can help researchers identify such biases and develop tools to mitigate them. However, AI systems have shortcomings of their own: they can hallucinate, make mistakes and are often designed to please users.
"AI wants us to keep using it. It doesn't want us to abandon it. That's partly where its tendency to flatter users comes from. If it constantly argued with us, we probably wouldn't want to interact with it," he said.
Suvorov warned that as AI becomes more accurate, people may become overly reliant on it.
"It makes fewer mistakes than before, but that creates a more fundamental risk — the temptation to hand over increasingly complex tasks and stop paying attention ourselves," he said.
The St. Petersburg International Economic Forum is taking place from June 3 to 6.
