Google shares drop $100B because of Bard | Hacker News Reader

Google shares drop $100B because of Bard

npr.org

12 pointsnigamanth8 comments

This article is from Thursday morning.

Google's shares continued to drop through the end of the week, and the final loss of value looks more like $200 billion.

Read a few news articles that it was due to an error in their bard ad... like okay, nobody is claiming these things are always correct... now if you read their ai strategy blog post, you'll find the narrative was underwhelming. Akin to "we're adding ai to hangout!" And like five other dumb examples. Seemed out of touch with what's happening. I could see that costing billions in value.

Deepmind is the fast boat. Google is the boat anchor stuck underwater holding back Deepmind's Ai.

And probably by design. maybe we'd have chatgpt 10 years ago. Too bad they couldn't acquire openai!

I reiterate; https://news.ycombinator.com/item?id=34497650

Well, Google did say 'Bard is an experimental conversational AI service'. Not even ChatGPT or Bing AI are any more accurate or reliable and both are also experimental and in preview. [0]

I can confidently say that this stock drop is a typical over-reaction from the markets. Today, neither AI is ready to even challenge or replace search engines due to their hallucination issues and LLMs like ChatGPT are unable to cite their sources or even explain them.

If you cannot trust the output from a chatbot or an AI to generate correct or factual information, then it has little chance in being useful enough to be a reliable search engine replacement.

[0] https://www.theverge.com/2023/2/7/23589536/microsoft-bing-ai...

I don't know that people are reacting to the error per se, so much as they're reacting to where they saw it: in a major, high-glitz public announcement. Like NPR quotes others, the human beings who prepared the announcement could've googled to make sure they didn't demo a product working as unintended.

I think people understand AI's current limitations. What they've lost confidence in is Google's judgement, not Bard's.

Some have argued that Reddit was doing a cleanse of anything advertiser-unfriendly (quarantining subs, zero tolerance towards violence/anti-lgbtq+, etc) for its IPO. After these events I’d argue that OpenAI was paying them huge lumps of $$$$ to clean up the dataset that was training their LLM. Why not let unpaid moderators do the dirty work of classification and labeling?

Reply on news.ycombinator.com