An SEO’s guide to ChatGPT prompts

For weeks, marketers have been hearing about ChatGPT and its applications for content and SEO. 

Prompt engineer is even now a job:

I won’t pretend to know which SEO and digital marketing functions ChatGPT and AI software will and won’t replace. But becoming an expert at creating prompts for these tools can already be a valuable skill (and will likely increase in value over time as the tools improve).

This article will discuss what to keep in mind when creating ChatGPT prompts for SEO and share an extensive list of SEO-focused prompts to use in your day-to-day work.

ChatGPT favorite SEO prompts

ChatGPT prompt engineering strategy

Before getting into specifics, it is helpful to have a general approach to ChatGPT prompts (and AI chat / writing prompts) so that you can create prompts for your specific applications and be aware of great SEO prompts other folks have come up with.

ChatGPT prompt engineering strategy

First, it’s vital to understand ChatGPT’s limitations:

  • ChatGPT is trained on a large data set and finished training in early 2022, meaning not all of the information is up to date and – as you likely know as a user of the internet – not everything it’s trained on will be correct.
  • ChatGPT does not crawl the web! It has access to information about the web from when it was trained, but it will not go to a page and crawl it (but often, it’s responses will lead you to believe it did or does!)
  • There are a variety of things ChatGPT can get flatly wrong: facts, math problems, code.
  • While it has access to a lot of data and can organize that data in interesting ways, it’s unlikely to have an expert-level of “taste” or knowledge about any individual subject (yet, anyway).
  • ChatGPT wasn’t designed to be an SEO tool, so while you can create prompts for things like keyword research, clustering, link building, etc. keep in mind that it’s responses and suggestions frequently won’t be using the same types of data sets (driven by search popularity and competition) as your favorite SEO tools.

So if you ask ChatGPT for a list, don’t expect it to be curated at an elite level. If you ask it to write code, don’t just assume it will work.

If you ask it to write an article, don’t assume everything it generates will be accurate or well-written (particularly on topics that would require up-to-date information). 

QA is your friend!

I like to think of prompts kind of like I think of tasks like creating a content brief, or using a search operator, or creating an SOW. 

Based on that approach, here are specific tips for crafting a good prompt:

  • Be as clear and thorough with your prompt as possible.
  • Make sure ChatGPT is capable of doing what you’re asking it to.
  • Refine your prompt as you see output, and use your experiences with ChatGPT to inform future prompts.
  • Try to anticipate and address the areas where it may get tripped up as much as possible with the prompt itself.

The best (and worst) ChatGPT SEO prompts

As with most topics, many people are sharing great information about ChatGPT prompts you can use for SEO. And then there are some sharing prompts which are somewhere between mediocre and downright harmful.

Below, I listed examples of what I think are largely “good” (or helpful, to steal a term from Google) prompts and what I think are “bad” (unhelpful to harmful) prompts.

These examples offer specific ideas for prompts that can help make you more efficient, and also some prompts that can get you into trouble (or waste your time).

The good

We’ll start with some useful prompts. Again, I can’t emphasize this enough: QA the output of everything you get from ChatGPT! 

Information can be (dangerously) wrong or misleading and code can break in the worst ways (more on that later in the article).

While Google has stated AI-generated content isn’t explicitly in violation of their search guidelines, it’s certainly possible to get your site into trouble with it.

Keyword and topic brainstorms

Due to the possible downside above, I like to think of ChatGPT as a helpful brainstorming tool for various tasks, including keyword research.

It is particularly useful as an early starting point for getting ideas for keywords and topics.

Let’s assume I’m creating a site aimed at delivering information about coaching for youth basketball coaches:

chatgpt topic idea brainstorm

As someone who is frequently looking for specific youth basketball drills I can tell you that this is a pretty good starting point.

Adding the note about popular websites and not just asking for ideas got me an extra layer of both topics and some examples of suitable sites in the niche. (Again, remember, these are from early 2022 or before!)

Let’s dig a little deeper into keywords:

chatgpt topics and keywords brainstorm

OK, interesting. I wonder how ChatGPT is determining that these are “popular and low competition.”

ChatGPT data sources and keywords

Wow, ChatGPT uses Ahrefs, Moz and Keyword Planner to generate keyword ideas? It only took a few seconds too! 

I felt like my LinkedIn and Twitter timelines would have been flooded with information about this, so I decided to ask a follow up:

ChatGPT wrong answer

So this is not what I asked. I’m getting a “Did you eat these cookies ChatGPT? Well you know the thing about cookies is…” vibe here, but let’s clarify:

ChatGPT clarifies wrong answer

Ah! (Much quicker to apologize than my kids, anyway).

Let’s take a quick look at whether these terms are actually low competition according to Ahrefs, one of the tools ChatGPT initially claimed to have used (tough break for Semrush and other tools here, incidentally):

ChatGPT suggestions - Ahrefs

Yeesh. Well, not much search volume here, but let’s check competition, particularly against this initial claim:

  • “SERP Analysis: I also did a manual search on Google for each keyword to see what kinds of results show up. If the top results are from high-authority websites with strong backlink profiles, it’s a good indication that the keyword is highly competitive. Conversely, if the top results are from low-authority websites with weaker backlink profiles, it’s a good indication that the keyword is less competitive.”
Ahrefs competition ChatGPT suggestions

This isn’t an impossible term competition-wise, though it is likely not feasible for my brand new site any time soon. I can see with the volume for other terms these pages are ranking for that there’s likely search volume in this area generally. 

A popular prompt framework is to get ChatGPT to answer a prompt “as an X” with X being a person with a specific job, level of experience or expertise, etc. 

Let’s see how that impacts things here:

persona prompt

These seem pretty good again! Let’s dig deeper – the combination of volume and competition maybe tripped the tool up. I’ll ask for terms with volume this time:

ChatGPT translate

And let’s see the data on that round via Ahrefs:

Ahrefs unpopular kewyords

Not much better! Keep in mind here that ChatGPT takes my “prompt history” into account for a specific chat, so if I’d opened a new chat and focused more on volume, I may have done a bit better.

Generally, it’s often the case that if you feel like you’re down a rabbit hole with ChatGPT and aren’t getting the data or responses you want, you may want to consider starting fresh and better orienting the tool to what you want.

These prompts and responses are a perfect encapsulation of ChatGPT for a lot of functions:

  • The initial set of results gave a pretty good framework for sites to check out and general topic areas
  • As I niched down and needed to develop specific, actionable plans with search-specific data things got murkier. 
  • Those terms don’t look great in terms of volume, my site can’t rank for those terms, and ChatGPT flat out told me something that wasn’t true about how it generated the suggestions! It also states everything very emphatically.

If I were looking for general topic ideas here, this might have been a useful exercise. 

If I had just started writing (or paying writers) to create content to target these phrases without checking ChatGPT’s suggestions, I might have wasted a ton of time and resources.

Title ideas

You should probably have your own ideas about creating titles and title tags for blog posts, but again ChatGPT can be a good place to generate some ideas:

ChatGPT

List ideas

Similarly list types like X examples, tips, quotes, etc. require some digging to come up with ideas – while you want to curate your own list you can quickly get ideas from ChatGPT:

ChatGPT list ideas

FAQ ideas

To quickly get FAQ ideas for an article the tool can be useful as well – again layering in specific instructions and a persona can help:

ChatGPT FAQ ideas

Content outlines

This is also true for content outlines. Again, keep in mind that tools like Clearscope, Content Harmony, Market Muse, Frase and Surfer SEO are creating briefs and outlines based specifically on what’s ranking in search results, while ChatGPT isn’t.

Here is the prompt I used:

  • “Create an outline for an article about teaching middle school players to move off the ball. The article should be for someone new to coaching basketball but with an understanding of the sport, and is aimed at beginning players. It should include things like: the types of offenses that encourage movement, specific examples of movements players should be making off the ball (cuts, picks, etc.) as well as some specific drills coaches can run to encourage movement.”

And here is the output:

ChatGPT content outlines

Again, not perfect, but a pretty decent framework! 

Keyword clustering

Standard caveats: Clustering is not going to be search-specific like tools like Keyword Insights or similar. It won’t necessarily be driven off of volume and competition, but you can create clusters either semantically or by things like levels of intent:

ChatGPT Keyword Clustering

And/or just add things like search intent as an additional column. 

Summarizing

Be careful with how you use these summaries (as they may get “flagged” as anything from low quality to AI over time). But if you have something like a study or a lengthy article that you want to feature in your own content, (and also let folks know you’ve featured in outreach) in something like a tips list, you can get help there:

ChatGPT summarizing

You can also ask for an outline of an article if you want to understand it quickly.

Technical SEO: Code snippets (schema, hreflang, etc.), robots, .htacess and more

Among the most dangerous ChatGPT prompts of all: code snippets! These can be great time-savers, but again: give as much detail as possible and QA, QA, QA!

A simple prompt along the lines of “Wrap FAQ schema around these questions and answers” will get you the code to copy:

ChatGPT code snippets

Same for different types of schema like organization schema:

ChatGPT organization schema

I have had various schema come back from ChatGPT that didn’t work when rendered on a live page, so again, be sure to check everything.

You can also have a robots file created:

ChatGPT robots file

And create rewrite rules. Be careful, though! (More on this later)

ChatGPT rewrite rule

Meta descriptions

A lot of SEOs or people responsible for maintaining lots of pages hate creating meta descriptions. This is really a perfect function for ChatGPT:

ChatGPT metadescription

Google Tag Manager, Google Analytics 4, Google Search Console and Looker Studio

You can get troubleshooting ideas for these platforms:

ChatGPT platform troubleshooting

Get instructions for building specific reports:

ChatGPT GA4 and Data Studio reports

Or even get code to use to interface with their APIs:

ChatGPT GSC API code

Translation

You can translate text to create country-specific pages:

ChatGPT translation

Formatting 

As an SEO, many quick and simple formatting tasks come up where ChatGPT can be very helpful. Things like converting a page to (or from) HTML:

ChatGPT HTML formatting

Extract links from a page:

ChatGPT extract links

(It missed the distinction between internal and external links here, obviously.)

You can also take a list of URLs and extract just the domains from them or convert a list of sites into HTML links (or vice versa).

Additionally, you can perform some functions you might typically perform in spreadsheets (which may or may not be more helpful than just performing them in spreadsheets, depending on your proficiency with spreadsheets and workflow).

Instead of concatenate or having to find a list of cities / copy and pasting a list of states:

ChatGPT spreadsheet functions

In place of VLOOKUP / IF types of functions, you can create a prompt like:

  • “Create lists of basketball drills from site A and site B, then show me A) the drills they both cover and B) the drills that are missing from each list that the other covers.”

Code for simple tools and widgets

As I walked through in this article, you can use ChatGPT to help you build simple tools like calculators that can enhance content, give you a chance to rank for specific terms (like subject calculator), and give you something to promote via outreach.

Rewrite content

If you give ChatGPT specific instructions on things like tone and what to include, it can help you rewrite or flesh out content:

ChatGPT rewrite content

Outreach assistant

While ChatGPT can’t find contact information for you (at least for now), there are some specific outreach tasks it can perform.

Like getting a list of ideas of places to guest post:

ChatGPT outreach assistant

Or drafting a template for an outreach email. This is potentially particularly valuable if English isn’t your first language:

ChatGPT outreach assistant template

Infuse your prompts here with your own tone and content preferences to make sure the template you get is closer to being consistent with your typical outreach emails.

You can even craft an entire auto-responder sequence:

ChatGPT Autoresponder Sequence

Or you could create different types of outreach lists:

ChatGPT outreach lists

And as you work through the outreach process, you can likely find even more opportunities to leverage ChatGPT.


Get the daily newsletter search marketers rely on.


The bad

In my experience, most of the trouble you’ll run into with ChatGPT will be:

  • Asking the tool to complete a process whole-cloth for you (rather than particular functions).
  • Not providing enough oversight.

Publications have already been accused of publishing AI-generated content with errors and plagiarized content. 

If we’d unleashed ChatGPT to do all of our keyword research for us, we likely wouldn’t have gotten a lot of traction.

When you try to assign it tasks like strategy, you’ll often get the same kind of boilerplate advice you’d get from a beginner-level X tips article on Google:

ChatGPT marketing plan

And without a lot of human input (specific prompts, editing, and likely a mix of human and AI content weaved together), you’ll likely get warmed-over content.

Depending on your purpose (and your risk threshold), that may be fine.

You may not need your meta descriptions, FAQs or certain articles or pages to be “10x”. But make sure you understand what you’re getting.

The ugly

Of course, some areas, like health-related content or critical tasks, on your site can go sideways:

Be particularly mindful of ChatGPT’s limitations 

When it comes to ChatGPT prompts and SEO, you want to proceed with caution.

Get good at creating your own prompts and sourcing inspiration (or productivity enhancements) from prompts other people share.

Here are a few great articles with additional SEO prompts for ChatGPT:

Opinions expressed in this article are those of the guest author and not necessarily Search Engine Land. Staff authors are listed here.

Is ChatGPT the future of cheating or the future of teaching?

ChatGPT, the chopping-edge chatbot from OpenAI that was unveiled in November 2022, can resolve math equations, generate a heritage term paper, compose a sonnet and virtually almost everything in amongst. So it is not shocking that lots of educators assistance banning the chatbot in faculties to prevent plagiarism, cheating and just plain inaccuracy.

In response to these considerations, some key districts have banned the chatbot in faculties. In December, the Los Angeles Unified College District “preemptively” blocked entry to ChatGPT although “a danger/profit evaluation is carried out,” a district spokesperson instructed the Washington Put up. And in January, New York Metropolis Public Schools banned accessibility to ChatGPT from units and networks that the school owns, per the Washington Post. A spokesperson for the NYC Office of Training instructed Chalkbeat that the choice was created “due to worries about destructive impacts on university student discovering and considerations with regards to the safety and precision of articles.”

But not everybody is on board with a total ban — some in the schooling globe say instead of banning it, educate little ones how to use it well and quite, and it could be a beneficial academic device.

A person matter that specialists do concur on: Do not just dismiss it.

“ChatGPT is just just one far more piece in the puzzle that’s accelerating transform in conditions of training and instruction,” explained Trevor MacKenzie, an English trainer in Victoria, Canada.

The largest problem with ChatGPT: cheating

ChatGPT is the primary edge of a new wave of chatbots based on so-identified as generative language products. They are experienced on massive sample sets of composing, and their crafting is considerably far more structurally complex than prior generations of bots.

The chatbot is so good at building human-like responses that Alex Lawrence, professor at Weber Point out University, explained it as “the best dishonest instrument at any time invented,” the Wall Avenue Journal reported.

The concern from some teachers is that they may possibly not be ready to differentiate among a student’s do the job and the get the job done of ChatGPT.

It would look like the worry is warranted. As CNN reported, ChatGPT is potent more than enough to have handed law tests in four courses at the College of Minnesota and a check at the College of Pennsylvania Wharton Faculty of Small business. It’s well worth noting, nevertheless, that it does from time to time stumble also. ChatGPT’s attempt at “basic equations of rocketry,” for case in point, was unsuccessful, NPR noted.

And according to study knowledge from Study.com of 1000 students aged 18 and up, additional than 89 per cent of pupils who responded said they had made use of ChatGPT to support with a research assignment, 48 per cent admitted they had utilized it for an at-residence check, and 53 {4224f0a76978c4d6828175c7edfc499fc862aa95a2f708cd5006c57745b2aaca} had utilized it to write an essay.

Lecturers faced a comparable situation when Wikipedia was introduced. The anxiety, then, like the fear now, is that learners would plagiarize info from Wikipedia alternatively of sourcing the data themselves.

Not like Wikipedia, there is no supply attribution with ChatGPT, which provides a further concern.

One particular of the most worrying items about ChatGPT, in accordance to Nadav Ziv, research affiliate with the Stanford Heritage Education Team, is that it makes info devoid of citing resources. He explained it as “free-standing details,” or facts that is not attributable to a resource.

As a examination, Ziv requested ChatGPT to write an essay on the causes of the American Revolution, which it did, but there were being no sources attached to the details. Background can be subjective with scholarly disagreement, so the fret, he said, is that with no attribution, pupils could possibly address the data as truth without having taking into thought how the resource can impression trustworthiness.

An inevitable portion of the foreseeable future

Ziv thinks that banning ChatGPT, like banning Google or any useful resource on the net, is just about difficult to do. It’s not only that students will proceed to use these methods with or with no a ban, but it’s actually vital that college students know how to properly interact with technology like ChatGPT. “I consider it’s more about getting ready college students to safely and proficiently have interaction with ChatGPT,” he explained, “understand its limits, fully grasp its opportunities.”

Yotam Ophir, professor of communication at the College at Buffalo, pointed out a recurring sample with the introduction of new engineering: Each individual time a new technologies is launched, we obtain ourselves having difficulties with how it forces individuals to rethink the issues they do. The most effective comparison, he said, is calculators, which, like ChatGPT, numerous located threatening to education and learning. The fret, he defined, was about the risk of calculators and statistical application finally changing mathematicians.

The introduction of these new systems, he reported, can be disruptive in nature and spark a rethinking of values, which is from time to time essential.

“ChatGPT will be a component of their [students’] life,” he claimed. “There is no heading again.”

So as an alternative of prohibiting pupils from applying it, Ophir mentioned, educators ought to consider the reason of training in the to start with put. “If the intent is to put together children for the earth of tomorrow, and I believe it ought to be,” he said, “then ChatGPT is going to be a element of their lives.”

A advantageous instructional resource

There are a handful of distinctive methods to effectively integrate ChatGPT into new curriculums, say some specialists.

Kathy Hirsh-Pasek, professor of psychology at Temple College and a senior fellow at the Brookings Institution, defined that when pupils master to generate, the most essential techniques they are genuinely creating are getting equipped to produce a potent thesis from the details they have collected and then using paragraphs to support that thesis.

This ability of figuring out the ideal proof to use, she argued, would continue to be in the purview of individuals to establish. This are unable to be changed by technologies. In the upcoming, Hirsh-Pasek thinks, we will price curiosity, significant considering and creative imagination over rote memorization, and ChatGPT will enable foster these expertise.

“We in fact aid learners a lot more by displaying them how to request the appropriate issues, which I assume is an significant talent,” she reported, incorporating that she encourages students to use their “mental muscles” to make what they are performing on greater than what they received from ChatGPT.

One more benefit of ChatGPT? Supporting teachers rethink how they educate.

In accordance to MacKenzie, ChatGPT forces teachers to change their instructional tactics and accelerate change, enabling lecturers to replicate not just on what they educate but how they train.

“If lecturers teach from a stance that simply will allow understanding to be a sequence of transactions (assign, transform in, repeat),” he discussed, “then ChatGPT forces instructors to change to a more system-targeted teaching stance where we more over and above the transactional and into a room of authentically acquiring competencies.”

And then there is also a way for ChatGPT to be made use of to instruct learners about intellectual house and plagiarism, according to Olga Polites, chief of the New Jersey chapter of Media Literacy Now and a trainer with 40 years of encounter.

The point that ChatGPT doesn’t cite sources for the data it generates can be applied to teach pupils about their moral obligation to be the originator of the material that they are developing. “This is what ChatGPT does,” she mentioned, “it phone calls into query whether or not you are the originator.”

How teachers steer clear of the ChatGPT pitfalls

Indeed, there are benefits to incorporating ChatGPT into the classroom, but how do teachers do it properly and prevent the prospective for cheating and misinformation?

MacKenzie believes it all begins with a sturdy formative evaluation apply, which, as he clarifies, is furnishing coaching and comments in the course of the approach of a scholar acquiring a process or an assignment.

So alternatively of basically searching at the last product or service without offering opinions in the method, a formative assessment signifies a teacher is involved in the entirety of the course of action of the assignment becoming done. It would be difficult to cheat applying ChatGPT because, in concept, with this kind of assessment, practically nothing a pupil turns in would be a surprise to the trainer — it would be simple to distinguish between the student’s get the job done and the get the job done of ChatGPT.

Equally, Hirsh-Pasek thinks the solution is in readjusting the way instructors educate.

“Part of this is going to be a pedagogical shift and a mind change,” she mentioned. The results of studying, she stated, have to have to be discovering how to be a lot more collaborative, learning how to make communities, creative thinking and resourceful innovation, and the self confidence to consider challenges even if they lead to failure. But all these competencies, she reported, are antithetical to the present school system. But ChatGPT, like covid, can be the motor for transform, she argues.

“Covid and ChatGPT are engines that are relocating us towards adjust,” she stated, “because they are so spectacular that they pointed out how frail we are.”

Thanks to Brett Zach for copy editing this article.

ChatGPT is confronting, but humans have always adapted to new technology

Adapting to technological improvements is a defining portion of 21st-century existence. But it’s not distinctive to us: it can be been portion of the human tale because our earliest published records – even showcasing in the plotlines of historical myths and legends.

When ChatGPT threatens to alter producing (and producing-associated work) as we know it, the Mesopotamians, who lived 4,000 several years back (in a geographical place centred in present day-working day Iraq), went by means of this sort of seismic improve in advance of us. Their civilisation is credited with the invention of crafting.

Residing in changing periods

Just two months after launching in November 2022, OpenAI’s ChatGPT has by now achieved an audience of in excess of 100 million people today.

The large language product (LLM), at times known as “autocomplete on steroids”, has drawn focus for its potential to build human-like responses to queries. Its operate has received passing grades on university law and enterprise exams, and been used in a Colombian court docket to help make your mind up a lawful ruling.

Responses to ChatGPT and its opponents have vacillated involving recognising the likely threats of the chatbot and extolling its attainable added benefits. We’re flooded with theories on how new developments in AI may transform how we function, examine and reside.

Ancient Mesopotamia was home to several of civilisation’s early developments. Its folks were being environment leaders in adapting to technological and cultural adjust.

Mesopotamians invented the wheel and agriculture, and pioneered innovations in mathematics, urbanisation and transportation. These breakthroughs are mirrored in cuneiform literature, one of the oldest regarded sorts of producing.

The background of cuneiform crafting is sophisticated, but it appears to be to have to begin with created to history economic details, these types of as money owed that had been owed. Over time, nonetheless, the Mesopotamians widened their use of signs inscribed into clay tablets to history a variety of details, in various languages. New makes use of coated all the things from diplomatic correspondences to omen texts, to some of the world’s oldest literary masterpieces.

Tech-assisted heroes

In the world’s earliest recognised composed epic, Gilgamesh, the eponymous hero is demonstrated inventing and utilizing technologies, these as diving weights and a sail, to additional his journey to the edges of the globe – and beyond.

As pointed out by Assyriologist Andrew George, the younger hero develops new technologies to support his quest for fame and immortality. These innovations make it possible for him to engage in earlier mysterious functions, these kinds of as sailing and deep-sea diving.

Another royal hero from Mesopotamia, Lugalbanda (often identified for his tremendous velocity) is also credited with technological progression. Lugalbanda increases the strategy of hearth-beginning by applying flint to spark embers and bake bread. The heroes’ use of new applications emphasises their exceptionalism.

Inventions and ambiguity

Mesopotamian epics really don’t present cultural and technological advances as unambiguously and uniformly effective. In Gilgamesh, the gains of civilisation and urbanisation, these types of as innovations in wall-making technological innovation, are juxtaposed with their prices – such as environmental destruction and alienation from the wild.

Certainly, the epics often signify new systems currently being harnessed in the service of human conflict – and disproportionately serving the pursuits of all those with significant social standing. In the Sumerian epic Enmerkar and the Lord of Aratta, the heroic king defeats his enemy by inventing and using remarkable technological know-how: the capability to publish on clay tablets.

This creation is also imagined to be referenced in the Epic of Sargon, wherever Sargon appears to dodge a murder try through his epic looking at skills. The text notes that while producing on tablets experienced been designed at that phase, the use of envelopes to conceal their contents experienced not (probably thankfully for Sargon).

In some approaches, the representation of new systems in cuneiform literature echoes present-day worries about AI: fears of raising social inequalities and its prospective use in cyberwarfare.

The long run of history

Finding out the past can deepen our comprehending of how human beings have tailored to present day technology above hundreds of several years. Conversely, modern technologies carries on to broaden our comprehension of historical past.

In latest decades, AI – the latest form of writing – has been applied to decipher the oldest: cuneiform literature. The Fragmentarium task, for example, employs sophisticated algorithms to establish which fragments of shattered cuneiform texts belong with each other these algorithms predict the textual content that after crammed the lacking sections.

AI will very likely proceed to change the way historians analyse the earlier. This will call for new things to consider all around acquainted challenges – this sort of as how to stand for the earlier properly in light of perhaps biased proof, and the will need to critically appraise sources of data.

In the broader industry of academia, the boundaries of how AI may well be used have not however been plainly defined. In January, for illustration, a leading worldwide AI meeting banned the use of AI equipment for composing scientific papers – while its use in enhancing papers was accepted.

Taking into consideration the limits of engineering

Even those early tech adapters, the Mesopotamians, ran into issues the technologies of the day could not handle.

Local weather alter is imagined to have resulted in the downfall of the Akkadian Empire, at times termed the world’s first multi-national political entity. And even the crafty Gilgamesh could not escape his have mortality.

Human beings have been grappling with how to invent, use and adapt to technological know-how considering the fact that our earliest civilisations. In Mesopotamian epic literature, new know-how can help heroic individuals travel over and above accepted limits and acquire new abilities. But the engineering and resulting knowledge are not often evenly dispersed.

Knowing how we tailored to shifting technology in the past helps us a lot more entirely fully grasp the human ailment – and might even help us get ready for the upcoming. (The Discussion)

Melbourne, Feb 13 (The Conversation) For Valentine’s Working day, some couples only roll their eyes at every single other in mutual cynicism. The capitalisation of enjoy in the modern planet can definitely appear banal.

But Valentine’s Working day gifts are rarely a contemporary invention. People today have been celebrating the day and gifting like tokens for hundreds of yrs.

We should first convert to Geoffrey Chaucer, the 14th-century poet, civil servant and keen European traveller. Chaucer’s poem from the 1380s, The Parliament of Fowls, is held to be the initial reference to February 14 as a working day about love.

This day was by now a feast working day of a number of mysterious early Roman martyred Saint Valentines, but Chaucer described it as a day for people to pick out their fans. He knew that was less difficult claimed than done.

The narrator of the poem is unsuccessful in really like, despairing that lifestyle is shorter in comparison with how extended it normally takes to discover to adore well. He falls asleep and desires of a back garden in which all the various birds of the environment have collected.

Mother nature describes to the assembled flocks that, like every 12 months on St Valentine’s Working day, they have occur to choose their partners in accordance with her regulations. But this method will cause confusion and discussion: the birds can’t agree what it implies to follow her principles simply because they all worth various factors in their partners.

Legal and emotional importance

Like right now, in Chaucer’s time gift-giving could be extremely ritualised and symbolise intention and commitment. In Previous and Center English, a “wed” was any sort of token pledged to assure a assure. It was not right until the 13th century that a “wedding” arrived to necessarily mean a nuptial ceremony.

The same interval observed marriage completely transform into a Christianised and unbreakable determination (a sacrament of the Church). New conventions of appreciate developed in songs, stories and other types of art.

These conventions influenced broader cultural ideas of emotion: adore letters ended up published, grand acts of services were celebrated, and tokens of adore had been offered.

Rings, brooches, girdles (belts), gloves, gauntlets (sleeves), kerchiefs or other personalised textiles, combs, mirrors, purses, bins, vessels and pictures – and even fish – are just some examples of romantic items recorded from the late middle ages.

In tales, gifts could be imbued with magical powers. In the 13th century, in a historical past of the planet, Rudolf von Ems recorded how Moses, when obliged to return house and go away his initial wife Tharbis, an Ethiopian princess, experienced two rings produced.

The a single he gave her would lead to Tharbis to neglect him. He constantly wore its pair which held her memory forever fresh new in his head.

Outside the house of stories, gifts could have authorized significance: marriage ceremony rings, critical from the 13th century, could show that a marriage experienced happened by evidencing the intention and consent of the giver and receiver.

The art of loving

Like Chaucer, 20th-century German psychologist Erich Fromm imagined men and women could understand the art of loving. Fromm assumed appreciate was an act of giving not just material points, but one’s joy, interest, knowing, awareness, humour and unhappiness.

Although these items could possibly consider some time and follow, there are far more simple ideas from historical past. Created playing cards have dominated given that the industrial revolution, taking their spot together with other now classic presents this kind of as bouquets, jewelry, intimate clothing and consumables (now more generally candies than fish). All can be personalised for that personal touch.

There have, of class, been weirder examples of like items, these types of as Angelina Jolie and Billy Bob Thornton exchanging necklaces with silver pendants smeared with each other’s blood.

Artist Dora Maar was so upset when her notoriously terrible lover Pablo Picasso complained about acquiring to trade a painting for a ruby ring she promptly threw the ring in the Seine. Picasso soon replaced it with a further, this a single featuring Maar’s portrait.

A superior really like token can long outlast the thoughts that prompt its giving: a flower pressed in a reserve, a trinket at the bottom of a box, a fading heartfelt card or a bittersweet tune that jolts you again to an previously time. In this way, the meaning of presents can alter as they grow to be reminders that all items pass.


Meet Bard, Google’s Answer to ChatGPT

Google is not about to let Microsoft or any individual else make a swipe for its look for crown without having a fight. The firm declared nowadays that it will roll out a chatbot named Bard “in the coming months.” The launch seems to be a response to ChatGPT, the sensationally popular artificial intelligence chatbot produced by startup OpenAI with funding from Microsoft.

Sundar Pichai, Google’s CEO, wrote in a blog article that Bard is presently obtainable to “trusted testers” and built to set the “breadth of the world’s knowledge” guiding a conversational interface. It utilizes a smaller variation of a effective AI model known as LaMDA, which Google initial introduced in Might 2021 and is based mostly on comparable technological know-how to ChatGPT. Google says this will make it possible for it to supply the chatbot to more consumers and assemble suggestions to aid deal with worries around the quality and precision of the chatbot’s responses.

Google and OpenAI are the two making their bots on textual content generation software that, even though eloquent, is prone to fabrication and can replicate unsavory models of speech picked up on-line. The require to mitigate these flaws, and the point that this sort of software package are unable to effortlessly be current with new data, poses a problem for hopes of building powerful and rewarding new merchandise on leading of the technology, including the recommendation that chatbots could reinvent world-wide-web research.

Notably, Pichai did not announce designs to combine Bard into the search box that powers Google’s gains. Alternatively he showcased a novel, and careful, use of the underlying AI technological innovation to boost typical research. For inquiries for which there is no single agreed-on answer, Google will synthesize a reaction that displays the differing thoughts.

For example, the question “Is it a lot easier to master the piano or the guitar?” would be fulfilled with “Some say the piano is much easier to discover, as the finger and hand actions are more organic … Other individuals say that it is less difficult to learn chords on the guitar.” Pichai also reported that Google designs to make the fundamental technological know-how readily available to builders by an API, as OpenAI is performing with ChatGPT, but did not provide a timeline.

The heady exhilaration inspired by ChatGPT has led to speculation that Google faces a severe problem to the dominance of its web look for for the first time in many years. Microsoft, which recently invested about $10 billion in OpenAI, is holding a media occasion tomorrow relevant to its work with ChatGPT’s creator that is thought to relate to new attributes for the company’s next-put search motor, Bing. OpenAI’s CEO Sam Altman tweeted a photograph of himself with Microsoft CEO Satya Nadella soon after Google’s announcement.

Quietly released by OpenAI previous November, ChatGPT has developed into an online sensation. Its ability to reply sophisticated queries with obvious coherence and clarity has numerous customers dreaming of a revolution in education, business, and each day existence. But some AI experts recommend caution, noting that the instrument does not understand the information it serves up and is inherently vulnerable to creating issues up.  

The situation may be notably vexing to some of Google’s AI gurus, mainly because the company’s researchers designed some of the technologies guiding ChatGPT—a truth that Pichai alluded to in Google’s site put up. “We re-oriented the enterprise all around AI 6 yrs back,” Pichai wrote. “Since then we’ve ongoing to make investments in AI throughout the board.” He title-checked equally Google’s AI research division and operate at DeepMind, the Uk-primarily based AI startup that Google acquired in 2014.

ChatGPT is designed on top rated of GPT, an AI design acknowledged as a transformer initial invented at Google that normally takes a string of textual content and predicts what arrives future. OpenAI has attained prominence for publicly demonstrating how feeding enormous amounts of information into transformer styles and ramping up the computer electrical power running them can create methods adept at generating language or imagery. ChatGPT enhances on GPT by having individuals provide opinions to diverse responses to an additional AI product that good-tunes the output. 

Google has, by its possess admission, chosen to carry on cautiously when it will come to including the technological innovation behind LaMDA to products. Besides hallucinating incorrect information and facts, AI types experienced on textual content scraped from the Web are prone to exhibiting racial and gender biases and repeating hateful language.

Those limitations have been highlighted by Google scientists in a 2020 draft analysis paper arguing for warning with text technology engineering that irked some executives and led to the company firing two notable ethical AI scientists, Timnit Gebru and Margaret Mitchell.

Other Google researchers who labored on the technology powering LaMDA grew to become disappointed by Google’s hesitancy, and still left the firm to make startups harnessing the exact same engineering. The arrival of ChatGPT appears to have influenced the organization to accelerate its timeline for pushing text technology capabilities into its solutions.

How to Block ChatGPT From Using Your Website Content

There is concern about the lack of an easy way to opt out of owning one’s material used to coach substantial language types (LLMs) like ChatGPT. There is a way to do it, but it is neither simple nor assured to get the job done.

How AIs Learn From Your Material

Big Language Types (LLMs) are properly trained on knowledge that originates from a number of sources. Several of these datasets are open supply and are freely made use of for schooling AIs.

In common, Massive Language Versions use a huge wide variety of sources to teach from.

Illustrations of the sorts of sources utilized:

  • Wikipedia
  • Govt court docket documents
  • Textbooks
  • Emails
  • Crawled internet sites

There are basically portals and websites providing datasets that are offering absent broad amounts of details.

One particular of the portals is hosted by Amazon, offering 1000’s of datasets at the Registry of Open Information on AWS.

How to Block ChatGPT From Using Your Website ContentScreenshot from Amazon, January 2023

The Amazon portal with 1000’s of datasets is just one portal out of a lot of some others that contain a lot more datasets.

Wikipedia lists 28 portals for downloading datasets, such as the Google Dataset and the Hugging Deal with portals for acquiring thousands of datasets.

Datasets Used to Teach ChatGPT

ChatGPT is based on GPT-3.5, also known as InstructGPT.

The datasets made use of to teach GPT-3.5 are the exact utilised for GPT-3. The big variance amongst the two is that GPT-3.5 utilised a approach identified as reinforcement studying from human feed-back (RLHF).

The five datasets made use of to train GPT-3 (and GPT-3.5) are explained on web page 9 of the study paper, Language Types are Couple-Shot Learners  (PDF)

The datasets are:

  1. Widespread Crawl (filtered)
  2. WebText2
  3. Textbooks1
  4. Publications2
  5. Wikipedia

Of the 5 datasets, the two that are based on a crawl of the Online are:

About the WebText2 Dataset

WebText2 is a personal OpenAI dataset made by crawling backlinks from Reddit that had 3 upvotes.

The thought is that these URLs are trustworthy and will contain high-quality material.

WebText2 is an extended variation of the authentic WebText dataset produced by OpenAI.

The first WebText dataset experienced about 15 billion tokens. WebText was used to train GPT-2.

WebText2 is slightly larger sized at 19 billion tokens. WebText2 is what was made use of to practice GPT-3 and GPT-3.5

OpenWebText2

WebText2 (developed by OpenAI) is not publicly available.

Having said that, there is a publicly accessible open-supply edition termed OpenWebText2.  OpenWebText2 is a public dataset developed making use of the same crawl patterns that presumably present identical, if not the similar, dataset of URLs as the OpenAI WebText2.

I only mention this in circumstance somebody would like to know what’s in WebText2. 1 can down load OpenWebText2 to get an plan of the URLs contained in it.

A cleaned up variation of OpenWebText2 can be downloaded in this article. The uncooked edition of OpenWebText2 is offered right here.

I could not uncover information about the consumer agent used for either crawler, perhaps it’s just identified as Python, I’m not positive.

So as much as I know, there is no user agent to block, though I’m not 100{4224f0a76978c4d6828175c7edfc499fc862aa95a2f708cd5006c57745b2aaca} selected.

Yet, we do know that if your internet site is linked from Reddit with at least a few upvotes then there is a great possibility that your web site is in equally the shut-supply OpenAI WebText2 dataset and the open up-supply variation of it, OpenWebText2.

Much more data about OpenWebText2 is in this article.

Prevalent Crawl

A single of the most usually utilised datasets consisting of Online material is the Typical Crawl dataset that is created by a non-financial gain firm termed Typical Crawl.

Prevalent Crawl facts arrives from a bot that crawls the whole Net.

The information is downloaded by companies wishing to use the info and then cleaned of spammy internet sites, etc.

The title of the Widespread Crawl bot is, CCBot.

CCBot obeys the robots.txt protocol so it is attainable to block Common Crawl with Robots.txt and stop your website knowledge from building it into a different dataset.

On the other hand, if your web page has previously been crawled then it is possible currently integrated in numerous datasets.

However, by blocking Common Crawl it is probable to choose out your web page information from becoming integrated in new datasets sourced from more recent Popular Crawl datasets.

This is what I intended at the pretty commencing of the report when I wrote that the procedure is “neither straightforward nor guaranteed to function.”

The CCBot User-Agent string is:

CCBot/2.

Increase the pursuing to your robots.txt file to block the Popular Crawl bot:

Person-agent: CCBot
Disallow: /

An more way to ensure if a CCBot person agent is legit is that it crawls from Amazon AWS IP addresses.

CCBot also obeys the nofollow robots meta tag directives.

Use this in your robots meta tag:

A Consideration Just before You Block any Bots

Quite a few datasets, together with Popular Crawl, could be utilised by providers that filter and categorize URLs in order to make lists of web-sites to focus on with advertising.

For instance, a firm named Alpha Quantum offers a dataset of URLs categorized applying the Interactive Advertising and marketing Bureau Taxonomy. The dataset is helpful for AdTech marketing and contextual promotion.  Exclusion from a database like that could bring about a publisher to shed potential advertisers.

Blocking AI From Employing Your Information

Look for engines allow internet sites to opt out of getting crawled. Common Crawl also allows opting out. But there is at the moment no way to get rid of one’s web page content from present datasets.

In addition, investigation experts really don’t seem to be to present web site publishers a way to decide out of becoming crawled.

The report, Is ChatGPT Use Of World wide web Information Good? explores the subject of whether it’s even ethical to use website details with out permission or a way to opt out.

Several publishers may perhaps take pleasure in it if in the close to potential, they are provided a lot more say on how their articles is employed, particularly by AI solutions like ChatGPT.

Whether or not that will materialize is mysterious at this time.

Much more assets:

Featured image by Shutterstock/ViDI Studio

How ChatGPT Could Take Microsoft’s Search Engine Bing Into the Future

Have you ever found yourself trawling through endless pages of results on a search engine to find the answer to a complex question? Say you want to find out if a vegetarian diet is suitable for your dog. Your research journey might begin by hopping onto Google and typing “is a veg diet good for dogs” into the search box and then having to make sense of the legion of generated links. By the time you find an answer, you’ve sunk way more time than you’d budgeted into poring through articles, reports and their sources. 

In the not-so-distant future, finding the answer to a complex question might not be such a tedious, mind-numbing process. Microsoft is reportedly integrating the AI tech that underlies ChatGPT into its Bing search engine in a move that could transform search as we know it. More specifically, Bing might have the potential to serve up a search experience that’s superior to Google, according to AI researchers, and potentially usurp the search giant’s decades-long dominance.

“ChatGPT is the first new technology in more than a decade that may really transform search and that could, at least in principle, upend Google’s market dominance,” said Anton Korinek, an AI researcher and professor of economics, at the University of Virginia. “What the technology does is that it allows consumers to interact with their computer in a much more natural and conversational form than traditional search.”

Read More: Why ChatGPT Will Be Everywhere in 2023

At this point, we don’t know what Bing’s AI-driven search results might look like exactly. Microsoft declined to comment for this story. However, AI researchers expect a meaningful departure from the status quo in terms of how a search engine presents an answer and how users interact with it. After all, ChatGPT is not designed to browse the internet for information (like a search engine). Instead, the chatbot uses information studied from vast swaths of training data to generate a response.

“ChatGPT can answer its users with a single clear response compared to the myriads of links of traditional search engines. It also has capabilities that are far beyond traditional search engines, like [the ability] to generate new text, explain concepts, have a back-and-forth conversation between the user and the system, and so on,” said Korinek. “People still find emergent capabilities that even the creators of ChatGPT were not aware that the system had.” 

Microsoft CEO Satya Nadella and OpenAI CEO Sam Altman at the Microsoft campus

Microsoft CEO Satya Nadella and OpenAI CEO Sam Altman at the Microsoft campus in Redmond, Washington, in July 2019. 


Scott Eklund

Microsoft announced plans on Jan. 23 to invest more resources into OpenAI, the creator of ChatGPT, to the tune of $10 billion. The deal would help keep both companies at the cutting edge of what’s known as generative AI, a tech used in ChatGPT that can learn from copious amounts of data to create virtually any content format (text, images, music and so on) simply from a text prompt. 

Search is just one in a suite of consumer-facing products in Microsoft’s stable that could potentially change meaningfully for customers in the coming years. According to a report by The Information, the Seattle-based tech giant also has plans to integrate ChatGPT’s AI tech into long-established products like Word, PowerPoint and Outlook in an endeavor that could change how more than a billion people work and accomplish daily tasks. For instance, integrating it into Outlook could mean simply prompting the email application to write a message about a specific topic. 

“Microsoft will deploy OpenAI’s models across our consumer and enterprise products and introduce new categories of digital experiences built on OpenAI’s technology,” the company said in a press release announcing the expanded partnership.

Conversational search

For its part, Google and its cutting-edge subunit DeepMind have been working on similar systems for years. The search giant chose not to release them to the public, though, in part over concerns about unethical behavior and how chat systems sometimes break social norms. For instance, in 2016 Microsoft created a chatbot called Tay that it was forced to take offline after it spewed out hate speech. Even ChatGPT, which has rules to create positive and friendly content, can be manipulated into producing upsetting responses using the right prompts.

However, DeepMind CEO Demis Hassabis said in an interview with Time this month that his company is considering its own chatbot, called Sparrow, for a “private beta” sometime in 2023. Google also plans to demonstrate a version of its search engine with chatbot features as early as this year, according to a New York Times report citing sources familiar with the matter. 

It’s no secret that Google search has become more conversational in general over the years. The company has made progress in this area with the Google assistant and with knowledge panels in search, and for years has pitched conversation as the future of search, demoing its AI systems LaMDA and MUM at its 2021 I/O developer conference. 

Leveraging OpenAI’s artificial intelligence seems to be how Microsoft is attempting to edge out Google at its own game. In the wake of ChatGPT’s release, Google management issued a “code red,” according to The New York Times. The report said internal teams had been reassigned to kickstart work on AI between now and an expected company conference in May. 

Still, Google’s search engine today remains the undisputed market leader as it has for decades, commanding 84{4224f0a76978c4d6828175c7edfc499fc862aa95a2f708cd5006c57745b2aaca} of global search market share, compared to Bing’s 9{4224f0a76978c4d6828175c7edfc499fc862aa95a2f708cd5006c57745b2aaca} (although it has grown in recent years) in 2022, according to Statista.  

Google didn’t immediately respond to a request for comment on this story.

Read More: Microsoft’s New Tools Use AI to Generate Any Image You Imagine

A cardboard craft-style open laptop with a chatty robot on the screen.

Chatbots got a lot more complex with OpenAI’s ChatGPT tool.


Carol Yepes / Getty Images

How smart is ChatGPT?

As you’ve probably heard by now, ChatGPT is a sophisticated chatbot that went viral globally after its consumer release in late November as a free online tool accessible to anyone with an internet connection. The AI-powered chatbot made headlines thanks in part to its ability to churn out delightful poetry, generate meal plans and provide authoritative answers to complex questions within seconds after being prompted. The tech underlying it isn’t exactly brand new, but no chatbot had yet managed to capture mainstream fascination in the way that ChatGPT did. That’s largely because OpenAI built a snazzy user experience around the GPT-3.5 language model, and that’s the phenomenon we know as ChatGPT. 

GPT-3.5 is an improved version of GPT-3, which debuted in 2020 and which learned from vast tracts of data and code to help it achieve its abilities. According to researchers at Stanford University, GPT-3 was trained on 570 gigabytes of text and has 175 billion parameters. (Google’s Dale Markowitz, meanwhile, put it at 45 terabytes of text data, “including almost all of the public web.”) For comparison, its predecessor, GPT-2, was over 100 times smaller, at 1.5 billion parameters. 

“This increase in scale drastically changes the behavior of the model — GPT-3 is able to perform tasks it was not explicitly trained on, like translating sentences from English to French, with few to no training examples. This behavior was mostly absent in GPT-2,” researchers from Stanford‘s Institute for Human-Centered Artificial Intelligence wrote in a 2021 post.

“The current version of ChatGPT probably already knows more about the world than any individual human, and it can present that knowledge in digestible form,” said Korinek.

For all the promise ChatGPT holds, there are nearly as many limitations. Critics of ChatGPT say it’s not always clear where the chatbot is pulling information from, which can make it difficult for people to trust the results. Skeptics also point out that ChatGPT will always remain undermined by the imperfect nature of the data it was trained on, including biased information or misinformation. 

OpenAI has acknowledged the chatbot’s weaknesses in its current form. CEO Sam Altman said in a December post on Twitter that the product struggles with “robustness and truthfulness” and that it would be “a mistake to be relying on it for anything important right now.”

But don’t look for the AI bandwagon to slow down.

“There will be a number of new systems like ChatGPT that will enter the market in 2023, and the main implication of the resulting competition is that consumers will have more choice and, hopefully, better products for consumers,” added Korinek.

GPT-4, which is under development, is reported to have 100 trillion parameters. But a release is not expected to take place until OpenAI is “confident we can [release] it safely and responsibly,” Altman said in an interview with StrictlyVC in early January. 

Altman also attempted to manage expectations of that fourth iteration of GPT, the sophisticated language model that underpins ChatGPT, saying “we don’t have AGI.” AGI stands for artificial general intelligence, or a technology with its own emergent intelligence as opposed to relying on the deep learning models currently used by OpenAI. It’s the kind of intelligence that has been dramatized in science fiction stories for more than a century and was popularized in recent years by the award-winning dystopian show Westworld. 

“I think [AGI] is sort of what is expected of us,” Altman said in the same interview, adding that GPT-4 is “going to disappoint” people who hold out that hope.

Editors’ note: CNET is using an AI engine to create some personal finance explainers that are edited and fact-checked by our editors. For more, see this post.