• The Neuron
  • Posts
  • šŸ˜ŗ OpenAI finally released ChatGPT Search.

šŸ˜ŗ OpenAI finally released ChatGPT Search.

PLUS: We share our honest thoughts on ChatGPT Search...

Welcome, humans.

Want to see what happens when you ask GPT to ā€œgenerate the most weird thing you can come up withā€? Here ya go.

Thereā€™s all kind of weirdness in the threadā€”we wonā€™t spoil it for ya! šŸ˜‰ 

Hereā€™s what you need to know about AI today:

  • OpenAIā€™s ChatGPT Search is now publicly available.

  • Meta is also working on its own Google Search competitor.

  • Claude now has desktop apps on Windows and Macs.

  • Microsoft is once again delaying its Recall AI feature.

OpenAIā€™s Google competitor is here!

ChatGPT Search, OpenAIā€™s AI search engine, is now live for paid ChatGPT users and is set to roll out to the freebie crowd over the next few weeks.

Weā€™re just going to go ahead and say it: ChatGPT Search is the biggest launch since ChatGPT (see our early review from July).

ChatGPTā€™s Achillesā€™ heel compared to Google has always been that you couldnā€™t search for live info (that Bing plugin was a flop...).

Now you canā€¦ and itā€™s really good.

Sam Altman even said that ChatGPT Search has ā€œdoubled my usage over the past few weeksā€. Biased, yes, but still... wow.

To use SearchGPT, open ChatGPT and click the globe icon labeled ā€œSearch the webā€ in the chat bar. Or, set it as your go-to search engine in Chrome using this extension.

SearchGPT in action

At this point, we can confidently say that ChatGPT Search is a better way to search info than Google: you receive answers faster and lets you dig deeper without endlessly clicking blue links.

This doesnā€™t mean Googleā€™s out of the AI search gameā€”far from it. Theyā€™re just stuck dealing with PR messes and weird hallucination glitches.

We know many of you love Perplexity for AI-based searchā€¦ and so do we. ChatGPT Search has a similar vibe, with a few notable distinctions:

  • ChatGPT Search is faster than Perplexity (both free + Pro), and weā€™ve found it to be better for quick facts.

  • ChatGPT Search is better for currents eventsā€”itā€™s more up to date and has a lot of partnerships with publishers, though it still hallucinates sometimes.

  • Perplexity might still be better for in-depth research (ChatGPT Search doesnā€™t suggest follow-up questions).

Check it out here and let us know your thoughts by filling out ā€œWhat did you think of todayā€™s email?ā€ at the bottom of the newsletter.

P.S. Meta is working on a search engine too. Innovative or just copying the trend? Weā€™ll let you decide. Just donā€™t ask the Winklevii their opinion...

FROM OUR PARTNERS

Ever wonder how OpenAI manages billing for millions users? Itā€™s Metronome. 

Pricing AI products is hardā€“using seats doesnā€™t cover your costs consistently, and product capabilities are constantly evolving.

Luckily, thereā€™s Metronome

Metronome powers billing and pricing for the top names in AI. Weā€™re talking OpenAI, people. 

Want the TL;DR on what Metronome can do?

  • Launch new products + pricing in hours (not weeks).

  • Connect billing directly to your product.

  • Get real-time usage data that actually makes sense.

Sound good?

You can request a demo right here. Or join their upcoming webinar to see the product and learn from Dropbox's former Growth lead about avoiding pricing pitfalls that sink most AI companies.

Treats To Try.

try Project Concept!

  1. *Adobe's Project Concept allows you to take your photos, stock images, brand assets, and more and turn them into remixable mood boards, revolutionizing the concepting process. Transform regions, mix styles, and collaborate instantly. Sign up to get early access here!

  2. timeOS organizes your meeting notes and action items while you focus on the conversation.

  3. Aptitude conducts hundreds of customer interviews simultaneously and gives you the key insights and themes from their responses.

  4. Langtail lets you spot dangerous responses in your models, like toxic content or wrong information, using spreadsheet tests before your app goes live.

  5. Google Research released InkSight, an AI model that converts offline handwriting into online handwriting format.

*This is sponsored content. Advertise in The Neuron here.

FROM OUR PARTNERS

Metaviewā€™s hiring.studio is your new AI copilot to generate and refine interview questionsā€”for free

Want to up-level your interview process? With hiring.studio, Metaview fine-tuned the worldā€™s best LLM to give you AI superpowers in hiring. 

With this tool, you can:

  • Collaborate + design interview questions tailored to your recruiting.

  • Automatically create a rubric for what to look for in an excellent, good, or bad answer.

  • Listen to sample candidate answers in formats like STAR.

The tool is free, and you can give it a spin by clicking here. The absurd mode is ROFL. šŸ¤£ 

Around the Horn.

  • Anthropic released Mac and Windows apps for Claudeā€”download here.

  • Sam Altman said in a Reddit Q&A that ā€œwe have some very good releases coming later this yearā€, none of which will be called ā€œgpt-5ā€.

  • Google Maps is getting new AI features thatā€™ll let you ask questions like ā€œthings to do with friends tomorrow night.ā€

  • Microsoft is once again delaying ā€œRecallā€, itā€™s conversational AI feature that can see your Windows screen.

Intelligent Insights

  1. A new study found Spanish questions get significantly less accurate election information vs English (52% inaccuracy vs 43% for identical English questions) from top AI models.

  2. Meta is making a bold move to get its Llama AI into government systems.

  3. The Department of Defense tested Scylla, an AI-powered security system that can instantly spot threats that human guards might miss while actually reducing false alarms.

  4. AI writes clearer, more accessible summaries of research papers than scientists do, and readers trust scientists more when their work is explained by AI (though they don't know it's AI doing the explaining).

  5. According to Qualcomm's CEO, smartphone apps may gradually be replaced by AI agents that will serve as a unified interface across all devices.

  6. This fascinating paper on LLM hallucinations says models actually ā€œknowā€ more than they showā€”they often internally encode the right answer even while consistently giving wrong ones.

A Cat's Commentary.

Thatā€™s all for today, for more AI treats, check out our website.

The best way to support us is by checking out our sponsorsā€”todayā€™s are Metronome, Adobe, and Metaview.

See you cool cats on Twitter: @nonmayorpete & @noahedelman02

What'd you think of today's email?

Login or Subscribe to participate in polls.