Bridging AI Trust Gaps Through Education and Transparent Integration

Key Quotes

"what's happened is in ai especially in the us is remarkable this is a nation that invented a lot of these technologies but shortly after the technologies took off with the launch of chat gpt in 2022 a number of american businesses ran a remarkable pr campaign a very successful one to tell the whole world ai is dangerous ai is like nuclear weapons don't trust the companies themselves building this technology it was a brilliant very successful pr campaign that i think delivered publicity maybe fundraising maybe lobbying influenced types of goals for a handful of businesses but it has damaged the public trust in ai in a very unfortunate way"

Andrew Ng argues that a successful public relations campaign following the launch of ChatGPT has damaged public trust in AI. He explains that this campaign, driven by a few businesses, framed AI as dangerous, akin to nuclear weapons, which has unfortunately obscured the technology's potential benefits. Ng suggests this PR effort may have served publicity or lobbying goals for some companies but at the cost of public confidence.

"so i'm not super worried about job loss now where the rub is is there is maybe 5 of jobs that are in trouble i think the translators you know ai translations can be really good so there's a very small fraction of jobs where ai could do 100 of what the person does and i think those jobs are in trouble i worry about voice actors as well but it's a small set of jobs i know that is maybe not the best message to say you know well for a small number of you you will lose your job i think unfortunately we're seeing a little bit of that but for the vast majority of people it will be their embracing ai that they still need people to do 70 of that job and they'll you know probably make more money have more fun if they embrace ai"

Andrew Ng addresses concerns about job loss due to AI, stating he is not "super worried" about widespread unemployment. He identifies a small fraction of jobs, like translators and voice actors, that are at risk of being fully automated. However, Ng emphasizes that for the vast majority of people, embracing AI will augment their work, allowing them to perform more of their job, potentially leading to increased earnings and job satisfaction.

"we live in a very polluted information ecosystem as it relates to ai right now because when ai technology was relatively new few people understood it and so again a handful of businesses got away with saying almost anything and both traditional media and social media were ineffective at fact checking them which is why a lot of the narratives about ai maybe has a gem of truth in it you know like you know there there unfortunately will be a little bit of job displacement but it's been dramatically hyped up well beyond the reality or ai is remarkably powerful and useful that's the gem of truth but this idea that we'll have agi artificial general intelligence in a couple of years and they'll just do everything that any human could do or whatever agi means that's also just not true"

Andrew Ng describes the current information environment surrounding AI as "polluted," where a lack of understanding allowed some businesses to make unsubstantiated claims. He notes that both traditional and social media struggled to fact-check these narratives, leading to exaggerated portrayals of AI's capabilities and risks. Ng points out that while AI has genuine power and may cause some job displacement, the more sensational claims about imminent Artificial General Intelligence (AGI) are not currently true.

"i learned a very interesting lesson about trust there so when my team used ai to try to figure out which patients were at high risk of dying high risk of mortality we then started notifying the doctors you know to to reach out to consider their patients for end of life care for palliative care and so predictably we should have seen this coming the doctors we called up they said who are you and specifically who are you to tell me that my patient might die and then what happened next was we missed it we actually built an ai dashboard to explain to the doctors hey this patient has this lab result they had this study this is why we think they're high risk of mortality we built this very detailed patient by patient dashboard to try to explain to the doctors why we thought a particular patient was at high risk of mortality thinking they would look at it and then try to challenge the judgment and we completely missed it"

Andrew Ng recounts an experience at Stanford Hospital where an AI system identified patients at high risk of mortality for palliative care consideration. He explains that doctors initially reacted with skepticism, questioning the AI's authority to make such pronouncements. Ng highlights that their attempt to build trust by providing detailed, patient-specific data explanations on an AI dashboard was ineffective, as the doctors' primary concern was not the data itself but whether they should trust the source.

"i think everyone should learn to code and this will make individuals much more powerful and much more productive and i'm already seeing this on my teams and a number of silicon valley teams so take you know marketing i i assume there are a lot of people here that that are experts in marketing i'm already seeing that marketers that know how to code can do much more than marketers that don't on my team when a marketer wants to launch a pr campaign you know they don't have to wait for an engineer to build a website they build it themselves one of my team members we couldn't get a hold of physical dials right in time to to measure sentiment as people watch a video or whatever but so one of my marketers rolled a mobile app you know so that we could do dial testing with having users just indicate the sentiment on the mobile app that or at that she'll build and this is not a software engineer and it turns out that marketers that know how to do this are much more productive and they actually can get paid more and i think today many people still can't imagine when we empower people with ai um teach people using it in the deepest way the amount of new things people will be able to do the amount of new value people will create and the pay raises and the new jobs that will be and a top of coding because one of the most important skills for the future is the ability to tell a computer exactly what you want so the computer can do it for you"

Andrew Ng advocates for widespread coding education, asserting it empowers individuals and increases productivity. He provides examples from marketing teams, where those with coding skills can independently build websites for PR campaigns or develop mobile apps for sentiment testing, tasks that would otherwise require waiting for engineers. Ng believes that teaching people to code, and by extension to communicate effectively with computers, will unlock new capabilities, create value, and lead to better-paying jobs.