My first newsletter using Hey World is a roundup of the notes I take when I'm researching an upcoming podcast. My co-host and I try to keep the show within reasonable limits, so it's rare that it stretches over 45 minutes. There's no secret sauce to podcast success rates based on episode length, my my subjective opinion is that there are far too many long podcasts, articles, books and other SEO-focused mediums trying to game the system. I think a show is better when it's tight, not flabby.
Of course, this means our notes can get left behind in the throes of a great discussion. In fact, isn't that true chemistry, when you can lift your eyes from the page and engage with what people are saying? Then again, what a shame to waste the research! There are some great links in there that people may find useful.
So here are mine, edited for brevity, but hopefully useful to Big Tech Little Tech fans who want to keep the episode 26 fire burning 🔥
About Google's search dominance
Oberlo: As of February 2023, 93.37% of all search queries across all search engine providers are done through Google. Microsoft Bing has 2.8% (6.3% in the US). Yahoo has 1.1%. China’s Baidu has a market share of 0.45%. The fourth-most used search engine in the US is DuckDuckGo. In the US, 95.1% of all mobile search queries are handled by Google, as opposed to 80.8% on desktops and 82% on tablets.
So, what’s new with Google?
Google’s AI search runs, in part, on a new, underlying technical model called PaLM2. While it works much like Google’s old model, PaLM, Google says it’s better at language, reasoning and code, and can run more quickly.
Google also says its new AI search engine will not answer queries when it’s not confident about the trustworthiness of its sources or when it comes to certain subject matters, including medical dosage advice, information about self-harm, and developing news events.
What are the other search engines doing to catch up?
Microsoft Bing: One of its recent developments in AI is the introduction of an AI copilot in its Power Apps platform, which can generate code and help with application development.
It also made updates to the Bing search engine, leveraging AI to provide more relevant and personalised search results.
It introduced various AI capabilities in its Azure cloud computing platform, including machine learning and cognitive services.
It sticks to a specific set of data sources to minimise ‘hallucination’ (Wikipedia and other encyclopaedias).
OpenAI's Davinci is the most powerful and feature-rich of OpenAI’s four AI large language models, but suffers in processing speed compared to Babbage and Ada.
Anthropic's Claude is a large language model designed as a competitor to ChatGPT, which adds restrictions designed to limit harmful or inaccurate responses.
DuckAssist is a work in progress, and may not generate accurate answers 100% of the time.
Search engines are a fascinating beast. What would the internet be without them? Over time, they've perhaps shown their worst versions of themselves, particularly around pop-ups, sponsored ads, misinformation, etc. Innovation in AI may be the chicken soup we all need to make our searches much more pleasurable, attractive and useful.
Until next time, thanks for reading my notes. I'll keep doing this. Much love 💜