Table of Contents
  1. In this episode
  2. AI News
  3. 1)GPT-4 is here
  4. 2) Google's AI PaLM toolset
  5. 3) Maker Suite: Google AI soon in Docs and Gmail
  6. 4) Google's Medical AI event
  7. 5) Claude, a ChatGPT competitor from Google backed Anthropic
  8. Google algorithm news
  9. The March Core Update is live! Finally!
  10. More SEO tips and interesting information
  11. More interesting ChatGPT tips and uses
  12. Local SEO
  13. SEO Jobs
  14. Marie's News
Login

AI and SEO: GPT-4, PaLM API, Claude: March 17, 2023 - Episode 278 - light version

Looking for the premium members version? Log in here or become a member. 

Last week’s episode (277)  See all episodes

This is a crucial edition of newsletter! With a whirlwind of recent developments in AI and search, my aim is to distill the most essential information and help you make sense of it all. This episode will catch you up on the latest news, and hopefully to inspire you with the incredible possibilities that lie ahead.

From the groundbreaking GPT-4 to Google’s under-the-radar announcement of their AI chat toolset, the tech world is abuzz with excitement. Yet, surprisingly, there’s been little chatter about Google’s event showcasing their advances in AI for medicine or the limited release of Claude, their ChatGPT competitor, for select businesses which will all be radical changes for our industry and the world.

As the potential applications of AI continue to expand, the implications for our everyday lives are both exhilarating and, for some, disconcerting. There is so much happening and it is difficult to keep up or even understand, yet alone get our jobs done and answer the questions our clients have. 

In this episode:

  • GPT-4 is live (and so good)
  • Google’s PaLM toolset is announced – like the ChatGPT API, but with Google’s AI
  • Maker Suite – Google’s AI will be in Docs and Gmail soon
  • Google’s Medical event – The advances they are making with AI are mind blowing
  • Google Claude available to some
  • A new Google core update – finally!
  • New guidance from Google on using their AI (do not pretend to have expertise or experience!

For paid subscribers only – Innovation and ideas

  • Improving a site’s author bio to demonstrate the expertise a searcher would be looking for (with examples)
  • Using the ChatGPT API to moderate and rank comments

Become a paid member for the full version

Each week I’m now sharing practical examples of AI’s use in search and soon will be working with subscribers in a community so we can improve together. Once I get my community going, current newsletter subscribers will be able to switch their membership for the same $18 per month. (It’s been this price for so many years now!)

Join the free version

There is now a lot more in the free version. Everything except the Ideas and Innovation section plus office hours. (We are having office hours next Wednesday, March 22)

Stay up to date on the latest on SEO & AI News

I'll send you an email each week once I've published newsletter.

Everything I've found interesting this week. Unsubscribe any time. Powered by Kit

Our world is changing

This morning I asked ChatGPT for another word for “smartass”. It taught me a new word, wiseacre.

If I wanted to know more about this word, historically I’d Google it, to find that it used to mean “soothsayer”. 

google search for wiseacre

But just look at this answer ChatGPT-4 gave me. I asked,

tell me more about wiseacre

“Wiseacre” is a term used to describe someone who often makes sarcastic or cheeky remarks, usually with the intention of showcasing their wit or intelligence. The word is a combination of “wise” and the Middle Dutch word “eiker,” which means “acre” or “field.” Together, they create a term that implies someone is “wise in their own field,” but it has taken on a more negative connotation over time. A wiseacre is typically seen as a person who enjoys being a know-it-all or showing off their knowledge, sometimes to the point of being irritating or annoying to others.

It’s important to note that being a wiseacre can be both endearing and frustrating, depending on the context and the people involved. If you identify as a wiseacre, you may find that your humor and wit are appreciated by some people but might be seen as off-putting or arrogant by others. To maintain positive relationships, it’s essential to be aware of how your words and actions may be perceived and to adjust your behavior accordingly.

 

This definition is so much better and left me with more questions! Unlike using Google search, I can continue to converse:

tell me about a few people who were considered wiseacres before the word was used to describe sarcasm

Before “wiseacre” took on its modern sarcastic connotation, the term’s original Middle Dutch meaning of “wijssegger” was associated with soothsayers, prophets, or wise men. It is essential to note that the term “wiseacre” as it is understood today did not exist in its current form during the time of these historical figures. However, some individuals from history who might have been considered as wise men, soothsayers, or prophets include:

  • Nostradamus (1503-1566): A French astrologer, physician, and reputed seer famous for his book “Les Prophéties,” which contains cryptic and poetic quatrains allegedly predicting future events. His work has been interpreted and reinterpreted over the centuries, and he is often seen as a prophet, despite skepticism surrounding the accuracy of his prophecies.
  • Pythagoras (c. 570 – c. 495 BC): An ancient Greek philosopher and mathematician who founded the Pythagorean school of thought. He was considered a wise man due to his profound influence on philosophy, mathematics, and science. However, some of his beliefs, such as the concept of metempsychosis (the transmigration of souls), might be seen as mystical or prophetic.
  • Oracle of Delphi: The Pythia was a high priestess at the Temple of Apollo in Delphi, Greece, and she was considered the most powerful and respected oracle in the ancient world. Her prophecies and guidance were sought by kings, generals, and ordinary people, and she was seen as a conduit for divine wisdom.

Again, it is important to emphasize that these historical figures were not labeled as “wiseacres” in the modern sense of the word. The term “wiseacre” evolved over time to describe a sarcastic or cheeky person who pretends to have more wisdom or knowledge than they actually do, which does not accurately represent the individuals mentioned above.

GPT-4 is dramatically better and more accurate than GPT-3. It is far more accurate and has significantly reduced the issue of generating inaccurate or fabricated information, which was a concern with GPT-3.5.

It is really, really good.

We have just entered an era where anyone with access to a computer can gain in-depth knowledge about any subject in the world through AI.

And we have yet to see what Google, with DeepMind, a leading AI innovator will give us. With the rapid pace of advancements in generative AI we may soon have the ability to understand and generate images, video, voice, virtually anything digital. In any language. 

Just this week a paper came out about technology that will let us speak in our language and have it seamlessly translated into another. 

And then there are the robots.

Recalling Sundar Pichai’s statement that AI technology is more profound than fire or electricity, it is possible that we are on the brink of a transformative era in human history.A vivid, thought provoking image of Sundar Pichai as he thinks about AI being more profound than fire or electricity --v 5

This episode of the newsletter is different from usual. I’m now providing much more information for free because it’s crucial for the SEO community and business owners to understand what is happening right now. For my paying subscribers, I’m introducing a new section called “Ideas and Innovation Discussion,” where we’ll delve deeper into specific cases and explore them in more detail. This section will evolve to be as helpful as possible and eventually move to a community platform that I’ve been working on for months. It’s almost ready to launch. We need a space to share and discuss the changes we are about to go through.

AI News

1)GPT-4 is here

GPT-4 is OpenAI’s newest AI language model.The difference between GPT-4 and 3 is that it has “human-level” performance on many tasks, understands context more effectively and has improved safety features. 

I have been incredibly impressed with GPT-4.

They improved the model with RLHF (Reinforcement Learning from Human Feedback), a process that refines the AI’s behavior by incorporating valuable insights from human-generated feedback to better align with user intentions and safety guidelines.

Also announced was OpenAI Evals, an open-source software framework designed to evaluate AI models like GPT-4 by assessing their performance on various tasks. It allows developers to create custom benchmarks, track model performance across versions, and share benchmarks to collaboratively identify potential failure modes and improve AI systems.

ChatGPT Plus subscribers (that’s the $20 per month paid ChatGPT plan) can access GPT-4 with a usage cap, which may be adjusted depending on demand and system performance. The GPT-4 API has a cost of $0.03 per 1k prompt tokens and $0.06 per 1k completion tokens, with default rate limits of 40k tokens per minute and 200 requests per minute. There is a waitlist to join. I have seen people getting access already.

Fort those using ChatGPT, you’ll find that for now you need to select GPT-4 for each new chat as it defaults to GPT-3.

 

ChatGPT defaults to GPT3.5

They are also limiting how many responses per 4 hour period you can get (usually 200, but sometimes less) although I’ve been using it a lot and have yet to be limited.

I have played with GPT-4 non stop since it came out and am very impressed. I’ll share more in the ideas and innovation section.

OpenAI’s announcement on GPT-4

People are doing some wild things with it:

 

I built an entire webpage with GPT-4 the night it came out. I discussed Google’s recommendations for product reviews after sharing it with ChatGPT. We discussed which content to cover, and wrote the content together with user intent in mind. We figured out the bits that would be important to a user and put them at the top. ChatGPT wrote the html and it worked perfectly. I put the <img> html into MidJourney to get images.

 

This excites me and also bothers me because it will soon be incredibly easy for anyone to create really good websites at a very low cost. But imagine what we can do when we spend even more time and effort on this!

I have a whole bunch of ideas on how we can take the information in Google’s guidance and use it in ChatGPT to improve our sites in ways Google really should reward. I’ll be sharing much more on this as I continue to think on it. In the new community I envision a place where I’ll be able to brainstorm on this with members’ content.

2) Google's AI PaLM toolset

It’s surprising how the GPT-4 announcement has eclipsed Google’s remarkable announcement! Don’t miss out on this major news:

Google has introduced an API for its AI language model, PaLM (Pathways Language Model). Not only can PaLM generate and edit text, but it’s also being trained as a conversational bot.

Select developers can access the PaLM API in Private Preview today, and a waitlist will be available soon. 

Thanks to Natural Language prompts, Google promises that PaLM will help businesses create text, images, code, videos, audio, and more.  This is going to be huge and will allow us to do so many things. 

I think the PaLM API will radically change the world.

PaLM API documentation on Google's site

Google’s announcement about PaLM

As we see more and more things that AI can do, Google will give us the ability to use this technology for our businesses as well. IMO the PaLM API is the thing most of us should be paying attention to. 

3) Maker Suite: Google AI soon in Docs and Gmail

Another huge deal!

Soon, we will be seeing Google’s PaLM (Pathways Language Model) technology right in Gmail, Docs, Slides, Sheets, Meet, and Chat. 

Google is encouraging us to use it to write our first drafts. I am not joking.

AI Chat in G Docs and Gmail

A select group of trusted testers will gain access to AI-powered writing features in Docs and Gmail, making it easier to draft, reply, summarize, prioritize emails, and create documents. Generative AI will also assist in creating images, audio, and video in Slides, provide insights and analysis in Sheets, generate backgrounds and notes in Meet, and enable workflows in Chat. 

Google did not say when this will be widely available to the public.

 

4) Google's Medical AI event

To me, this was the most important announcement this week – even more important than GPT-4. The things Google is doing and also plan to do with AI in medicine are mind blowing. Here is a thread with everything you need to know:

There was so much to get excited about:

 

  • Google’s Med-PaLM 2 language model excels in medical licensing exams: There is so much potential here for diagnostics and treatment, assisting healthcare professionals. Google shared how doctors will still be vitally important as caring and documenting symptoms is something a human excels at.
  • AI-enhanced ultrasound imaging: This enables doctors to use smaller, more affordable devices for patient triaging, increasing accessibility to crucial diagnostic tools, without needing extensive training.
  • AI transforming cancer treatment: Google has partnered with various institutions to make advanced models accessible to real-world clinicians for improved patient care.
  • Google and Harvard Medical School collaboration: Aims to revolutionize medical education and enhance the quality of healthcare.
  • Reliable YouTube health content: Youtube now shows indicators to show content is created by licensed health professionals or borne from personal experience.
  • Google Search improvements for healthcare: Easier access to Medicaid enrollment info and low-cost community health clinics, empowering users to find vital resources.
  • Expanded low-cost healthcare provider search feature: Google launched a feature across the US, connecting users to affordable care options.
  • Google and ThroughLine partnership: Connects users with mental health and crisis helplines, providing essential support services.
  • Fitbit offering notifications for irregular heart rhythms: Enhanced health metrics empower users to take control of their health and wellness. This is IMO a HUGE deal. They talked extensively about how the data gathered by Fitbit could be analyzed by AI. There is so much potential here as AI can analyze it and find commonalities and connections that could be connected to disease or poor health. And then I envision us or our doctors using AI to determine the optimal path for good health.
  • Mama’s Hub app in Kenya: Built with Open Health Stack, the app provides pregnant women with essential information for clinicians, improving maternal care. The presenter said, “This tool I’m putting in a woman’s hands will give her the same exact privileges as someone living in NY.” Wow. Pregant women in Kenya can journal their symptoms and then any medical care worker they see who is using the app has access to the insights gained from them.
  • Collaboration with the WHO: Aims to create a more coordinated healthcare environment, ensuring interoperability, standards, and good governance in healthcare. They talked about “continuity of care”. There was hint that the blockchain could be important here – essentially our medical information in our own private wallet that can be accessed in whole or part, when we give permission. 

To me, it feels archaic that our access to medical care largely depends on our ability to provide our doctors with a personal medical history, combined with their own level of knowledge. Imagine when our doctors, or really any doctor we see can get access to all of the important history and all of the recent medical research and innovations and expert discussions.  

The future seems to be moving towards a system where we can record our history or maintain a health journal, allowing AI to detect patterns that may indicate potential issues or prompt us to seek medical attention. Doctors will likely have access to tools that assist them in making more precise diagnoses.

This technology could enable pharmaceutical companies to create medications specifically tailored to each individual’s health needs. However, I imagine a world where AI could eliminate the necessity for many medications. I’ve had some intriguing conversations with ChatGPT about my theories, and that was before GPT-4. I plan to share these ideas someday, but not just yet.

5) Claude, a ChatGPT competitor from Google backed Anthropic

Anthropic, a prominent AI research startup backed by Google, has recently unveiled Claude, a next-generation AI assistant poised to compete with ChatGPT. Claude is designed to be helpful, honest, and harmless, and can be accessed via chat interface and API. It excels at various conversational and text processing tasks while maintaining reliability and predictability.

Claude can assist with a range of tasks such as summarization, search, creative and collaborative writing, Q&A, coding, and more. Partners like Notion, Quora, and DuckDuckGo have already integrated Claude into their platforms.

Claude is different from other AI assistants in that it is less likely to produce harmful outputs, more conversational, and easier to steer towards the desired output. Users can also give directions on personality, tone and behavior.

There are two versions of Claude: Claude and Claude Instant. The former is a high-performance state-of-the-art model, while the latter is a lighter, less expensive, and faster option.

Partners like Quora, Juni Learning, Notion, DuckDuckGo, and Robin AI have incorporated Claude into their platforms for various purposes, such as answering questions, online tutoring, improving productivity, generating search answers, and evaluating legal contracts.

You can go here to request access. I just did!

From Anthropic: Introducing Claude

 

Bing AI news

Bing AI search is running GPT-4 and has been for a few weeks:

It is now available in the Edge sidebar

Google gives us new guidance on using their AI

(Whenever we finally get to use it publicly!)

It is most interesting to see that the documentation talks about not making misleading claims of expertise or capability in YMYL areas (e.g. health, finance, government services, or legal.)

Google's guidance on using their AI

Google algorithm news

The March Core Update is live! Finally!

 

I swear I did not know this update was coming when I tweeted this. I just had a gut feeling that it was time.

 

The update is quite impactful for many sites. 

Here are a couple of sites I monitor that are seeing nice improvements:

early improvement with March core update

There are a few with similar hits as well. Many sites seem to have movement before the update. We have seen this before. Most likely if your traffic has changed dramatically even if it starts before March 15, it is due to this update.

Quick tip: If you are not sure whether your changes are due to the core update or changes you made to your site or perhaps seasonal changes, check whether the losses are also see in traffic from Bing or other search engines.

Given we are only a few days in, I have not analyzed this update.

If you have been impacted:

  • First, wait. The update is still rolling out. It will take up 2 weeks to complete. Sometimes there is turbulence where sites will see improvements and then declines or vice versa. In my experience though, when sites see changes like I’ve pictured above with most updates they usually stick. In my experience, most sites hit by an update feel the impact within the first 24 hours. Sometimes there are days within the update where many sites see adjustments.
  • Know that core updates are not about breaking the guidelines. You will not recover from a core update hit by disavowing. It is not about unnatural links. Although I think that in many cases, core updates make it so that Google can rely even less on PageRank as a proxy for quality and relevance. In most cases, technical website fixes are not the answer. These updates are about Google improving how they determine which content is relevant and helpful.
  • Look closely at pages that Google elevated above yours. Consider their content and your content in the eyes of Google’s guidelines on creating helpful, reliable, people-first content.

I will soon be sharing more – in a live, ever updating course on using ChatGPT plus a knowledge of how Google’s algorithms work to improve the quality of our content.

If you were impacted and you do make changes to improve quality and helpfulness, you will need to wait until the next core update to see if Google considers those changes substantial enough to want to rank your content alongside what currently exists on the web.

Google’s documentation on core updates

Google’s documentation on creating helpful content

Some of the most important questions to consider (based on the behavior of the last few core updates):

  • Does the content provide original information, reporting, research, or analysis?
  • Does the content provide a substantial, complete, or comprehensive description of the topic?
  • Does the content provide insightful analysis or interesting information that is beyond the obvious?
  • If the content draws on other sources, does it avoid simply copying or rewriting those sources, and instead provide substantial additional value and originality?
  • Does the content provide substantial value when compared to other pages in search results?

And most likely, given that this is the first core update since Google told us about rewarding content that demonstrates experience, these questions are important too:

  • Is this content written by an expert or enthusiast who demonstrably knows the topic well?
  • If someone researched the site producing the content, would they come away with an impression that it is well-trusted or widely-recognized as an authority on its topic?

From 2021, but still relevant:

 

More SEO tips and interesting information

The Helpful Content classifier is sitewide

 

There is a left sidebar being tested. (Good place for Bard?)

 

Having issues with getting the correct sitename and favicon to display?

 

You can ignore spammy referral traffic

 

Video from Google on Connecting GSC and Looker (Data Studio)

 

SearchLove sounded really good

 

Lily Ray’s talk about AI and SEO

https://twitter.com/resiever/status/1635955445318705152?s=20

 

GA4 is designed to integrate seamlessly with AI 🤔

 

Koray Guber interview. I have not had time to digest this yet but it looks very good. Koray is one of the few people I know who have a deep understanding of Google’s use of deep learning to rank content.

More interesting ChatGPT tips and uses

https://twitter.com/jbrowder1/status/1635720431091974157?s=20

(note – the generative AI capabilities of GPT-4 which will allow us to use it to understand images are not yet publicly available.)

If all goes as planned, I should be on the 6pm CTV news tonight talking about upcoming wave of spam callers we are about to encounter. They put it off for a few days as the reporter has been trying to clone another reporter’s voice to demonstrate. Thankfully none of the providers of voice cloning will do this for him without the other reporter’s permission.

CTV reporter chatting with me about AI

Tip: Make sure all of your family members and friends have a way to identify themselves that can’t be discerned from information online. I have a password with mine. Also, if I call you asking for money, it’s probably not actually me. 😛

MidJourney v5 is live!

It is so good.

I had previously been using Playgroundai.com as it’s free. But I decided to bite the bullet and pay for MidJourney. It was an excellent decision.

 

Here’s how to get MidJourney on your own server if you’d like. Personally I really enjoy seeing what other people have made on the Newbie servers.

 

Other interesting news and information

What happened with the Silicon Valley Bank potentially could impact the SEO industry.

My understanding is that for now, companies have not lost their money and will be able to pay their employees because of the actions of the US government. I included this because I know that MHC had many clients that were startups. We would commonly get messages saying, “Hey, we just got funding, now we can afford SEO!” I was worried that more agencies would lose clients as a result of this incident. I still am a little worried, but not as much now that the Fed has stepped in.

Here is more interesting info that I wanted to write about but damn, there is too much this week:

 

 

Local SEO

There is a big affecting some Google Business Profiles – Your business is not visible to customers

You can now report Google LSA Reviews

SEO Jobs

Looking for a new SEO job? SEOjobs.com is a job board curated by real SEOs for SEOs. Take a look at five of the hottest SEO job listing this week (below) and sign up for the weekly job listing email only available at SEOjobs.com.

Marie's News

My plans seem to change by the minute. I think the direction is clearer now though.

Course & Community

Got the Traffic Drop Assessment course close to completed. The current plan is to make this course free. There will be a paid course that guides you through improving your site based on Google’s recommendations.

This week you saw a new section in newsletter – Ideas and Ideation. I realized that we can accomplish so much as a group. The SEO community I’ve been working on for months now should be ready very soon so that we can discuss these advances and grow together.

Office Hours next week 

Let’s do a video call on Wednesday. All paid newsletter subs are welcome. See this section in the paid version for a link to join. We can discuss whatever is topical including the core update or ChatGPT use. We may have time to review traffic drops or even improve some pages with ChatGPT. This will be recorded for members who cannot make it.

Consulting

I have closed my consulting calendar for now.

If you need SEO consulting, here are a few options:

  • Sign up for my waiting list when I start doing site reviews again.
  • Contact me for a recommendation to a professional. (Please know I’m only in email 1-2 times per week now so there may be a bit of a delay in my response.)

Garden update

Seedlings are coming along!

I suspect this episode has been exciting for many and overwhelming as well. The world is about to change so much. Please know that if you are reading this, you are way ahead of the rest of the world when it comes to understanding what is happening with AI and how we access information. More and more people will be looking to us for guidance soon.

Have a wonderful weekend!

Marie

Premium Members Login

Lost your Password?