When Can We Trust AI to Make Key Business Decisions?
Read the Portuguese translation →
Author: Harvard Business Review · Watch on YouTube
Video authored by the channel shown, not by IANIA; all rights belong to the creator. Transcript and summary generated automatically; the original-language content is preserved.
Original language: English
Original video: When Can We Trust AI to Make Key Business Decisions? · Channel: Harvard Business Review · 01:00 · Watch on YouTube
Topics: AI agents, AI evaluation and benchmarks, AI governance and regulation, AI safety
Summary
Section titled “Summary”The discussion asks when AI can be trusted to make consequential business decisions without people checking every output. The response says the answer depends on the domain and on both technological and psychological realities. AI hallucinations have declined substantially since ChatGPT launched in 2022, but public sentiment has not improved as much. Society will be ready to delegate consequential decisions when people recognize technological progress and their confidence catches up.
Key points
Section titled “Key points”- Trust in AI for consequential decisions depends on the domain.
- There is both a technological reality and a psychological reality to consider.
- The AI hallucination rate has declined materially since 2022.
- Public sentiment about hallucinations has not improved as much as the technology.
- People need to see and recognize the degree of technological improvement.
- Society will be ready to delegate consequential decisions when technological progress is reflected in human confidence.
Chapters
Section titled “Chapters”| Start | Chapter | Summary |
|---|---|---|
| 00:00 | When to trust AI decisions | The question is when people can trust AI with consequential decisions without checking every output. |
| 00:30 | Technology and public confidence | Although hallucinations have declined, public sentiment has not kept pace with the technological improvement. |
| 01:00 | Readiness to delegate | Society will be ready to give AI consequential decisions when technological progress is reflected in human psychology. |
Quotes
Section titled “Quotes”I think it depends on the domain. — 00:00
There’s a technological reality and then there’s a psychological reality. — 00:00
The hallucination factor with AI has gone down quite materially since 2022 — 00:00
The public sentiment on hallucination has not improved as much as the technology has improved. — 00:30
when both of those things happen, that is when society will be ready — 00:30
Concepts and entities
Section titled “Concepts and entities”hallucination, technological reality, psychological reality, ChatGPT
Full transcript (original language)
Section titled “Full transcript (original language)”00:00 You know, in my mind as AI is going to achieve these sort of remarkable, you know, step change innovations, at a certain point we have to trust the output, right? What’s your, you know, suggestion to when we can have a comfort level with AI making consequential decisions and we don’t have to check everything? >> I think it depends on the domain. There’s a technological reality and then there’s a psychological reality. I’ll give you a very simple example. The hallucination factor with AI has gone down quite materially since 2022 and
00:30 ChatGPT first came out. The public sentiment on hallucination has not improved as much as the technology has improved. >> First impressions. >> Right? And so, part of this is you will actually need to make sure that you see the the degree of improvement that happens technologically. We will need to make sure that that actually is effectively reflected in the human psychology of it. And when both of those things happen, that is when society will be ready to say, “I’m going to give it
01:00 I’m going to be ready to give it consequential decisions.”