Connect with us

SEO

Can Google Detect AI Generated Content?

Published

on

Can Google Detect AI Generated Content?

Just when we thought that we could use everything AI to write all of our content, developers went and dropped AI writing detectors on us, too.

The creators of ChatGPT themselves released one just a few weeks ago, amidst new Google statements and updates on the whole issue of generative AI and content. 

It’s looking hard out there for these writing tools—but can engineers really develop a way to detect whether text has been by AI writers and conversational chatbots? Can Google detect AI-generated content? And, should you be worried about penalization if you’ve been using these tools yourself?

Let’s get into what this all means for your content production, and what you can do to avoid getting hit by Google’s algorithm. 

Is AI Content Bad for SEO?

Yes, AI-written content can be bad for your SEO. 

Google has said multiple times in the past that purely AI-written content goes against its guidelines. We’ll get into why and how it can be a problem for you later on.

Google also has a long, long history of using and developing AI tech. So, it’s safe to say that they do have ways to tell if your articles are written by an AI tool. 

Why AI-Generated Content Goes Against Google’s Webmaster Guidelines

Is using AI assistance in writing explicitly prohibited by Google’s Webmaster Guidelines? No. But, there are guidelines that strongly condemn ‘spammy automatically-generated content.’ Specifically, they call out any article written by AI for the “primary purpose of manipulating ranking in search results,” which they state is a violation of our spam policies.

Google's guidelines on spammy, automatically-generated content

This is a mistake you can commit if you’re trying to churn out the most content simply to climb the ranks as fast as possible. 

So, it’s not that AI-generated content is, by default, something that Google will punish you for. But, if you’re just mass publishing content using AI, then it may ultimately hurt your rankings, rather than help. 

Danny Sullivan—a renowned SEO expert and representative of Google—made this clear in a past statement:

We haven’t said AI content is bad. We’ve said, pretty clearly, content written primarily for search engines rather than humans is the issue. That’s what we’re focused on. If someone fires up 100 humans to write content just to rank, or fires up a spinner, or a AI, same issue…

How Can Google Detect AI Content?

To understand how Google can detect AI content, it’s important to understand how AI writing tools work.

Jasper, ChatGPT, and other similar AI use a process called Natural Language Generation (NLG) to generate copies and responses. NLG uses algorithms, trained on a large corpus of data, to generate human-like text. 

Google on the other hand can use machine learning algorithms to learn the different signals in content—such as text structure, grammar, and syntax in the text. This will help it detect patterns in the content that are indicative of AI generative writing. 

In short, if it falls within a structure that is noticed by the algorithm, Google can flag it.

This is entirely possible because AI writing is missing the natural variation, nuance, and complexity found in human writing. Plus, AI writing tends to be repetitive, and the facts and data they present may also be outdated or misinformative. 

So, Can Google Detect AI Content?

Yes, even if it’s been run through your AI tools a couple of times. It can possibly even detect it if it sounds very well-written. 

It might not be noticeable while skimming through the text your tools generate, and it might even read naturally—and yet, it can be detected. 

That’s not to say that Google’s algorithm is 100% accurate, though. When asked in an interview if Google was able to automatically detect the difference between human and AI content, Google’s John Mueller responded “I can’t claim that. However, if the web spam team sees anything that is automatically generated, they will take action.” 

So, it’s clear that there is no foolproof way of detecting AI-generated content, at least not yet. As AI technology and machine learning continue to develop, it’s safe to say that they will develop a highly reliable way to detect and penalize it.Other AI Writing Detectors

As I mentioned earlier, there are several AI detection tools out there that can catch AI-generated content. 

The one developed by Open AI, which they’re temporarily calling “AI Text Classifier,” is just the latest in the long line of tools you can use. Originality AI is a premium tool that charges you 0.1 cents per credit and checks your content for both originality and AI. Another option is AI Content Detector from Writer, which you can use for free—though only for 1,500 characters at a time.  

I tried them both myself. I first asked ChatGPT to write me an article:

ChatGPT writing an example article

Then, I put it into the AI detector tools, without any edits:

ChatGPT's example article being evaluated by an AI detection tool

Here’s what AI Content Detector had to say:

AI Content Detector's evaluation of ChatGPT's example article

So, these tools seem pretty reliable in finding purely AI-written content.

Should You Stop Using AI for Your Content?

“I’m using an AI writing tool, should I be worried?”

A lot of people are asking this question now. After all, with the introduction of ChatGPT in late 2022, and the explosion of AI software adoption, we all saw the incredible capabilities of AI. Suddenly, we all had free access to a tool that could replace a lot of the legwork and money that usually goes into content writing. 

How could we not use this technology in our SEO strategies, when it was so good? 

And now, months later, many are wondering if their AI-written content puts them in danger of a Google penalty. 

So, should you be worried if you used an AI writing tool for your site? The answer is—it depends. 

The situation as I see it is this: when you use things like ChatGPT to make content, you either publish it as is, or edit it before doing so. 

The first puts you in a bad position because you’re putting out bad content. It will be downgraded by Google because it’s low-quality content that doesn’t provide a helpful, valuable experience for readers.

Why? Because tools like this only scrape and remake content that was fed to them. They do not provide anything novel, since you’re not adding your expertise, your thoughts, or your interpretations to the content. 

So the name of the game now is knowing how to use these tools efficiently, while still adding quality (and a human touch) to your content. Always offer that expertise to it that makes it valuable to readers.

How Does This Affect AI-Generated Content?

You might still be skeptical, and feel safer if you’re writing content purely without AI.

However, what you should be worried about is choosing the best content writing tools, and making sure to use them properly. After all, given how much time and money you can save with this in your toolbox, it’s worth taking the risk—with a few guidelines set in place.

Google even supports this (in a way), in their updated page on AI-generated content:

Google's updated page on AI-generated content:

On the same page, they also said this: “This said, it’s important to recognize that not all use of automation, including AI generation, is spam. Automation has long been used to generate helpful content […] AI has the ability to power new levels of expression and creativity, and to serve as a critical tool to help people create great content for the web.

This is in line with how we’ve always thought about empowering people with new technologies. We’ll continue taking this responsible approach, while also maintaining a high bar for information quality and the overall helpfulness of content on Search.

So no, you don’t have to stop using your tools. But you might have to revisit how you’ve been using them.

For starters, I suggest working with the right team. If the people using them are complete novices, then it might do your content more harm than good. 

Why? Because if they don’t know how to use these tools properly, then they might end up producing low-quality content that will, at best, be ignored by users (and at worst, get penalized by Google). Either way, the result is something you need to avoid.

Plus, if they keep churning out that kind of content on your site, you’re probably going to see a dip in your traffic and engagement.

On the other hand, if you’re able to train them to use these tools in such a way that creates great content for your audience with less time and effort spent, then it will become an extremely valuable asset for your site. In the right hands, it’s a powerful thing. 

So if you’re going to use an AI writing tool, invest some time in learning how to use them, and figure out what you should avoid doing with your content creation. 

What to Do When Using AI for Content Creation

Here are a few things to keep in mind:

  • Check the writing style, tone, flow, and grammar of the generated content. Nothing turns off a reader more than something that is poorly written—or sounds odd.
  • Use the tool to generate the bulk of the text. You could start with an outline, and ask for a paragraph or two for each point. This will make it easier for you to target as many relevant talking points as possible with less effort on your end.
  • Provide clear guidelines. Without proper instructions, AI tools can struggle to generate quality content. If you don’t do this, you’ll probably waste time trying to re-generate the content you’re looking for. 
  • Understand what prompts you can use. AI tools are programmed to respond a certain way to prompts, which can help you generate content in different formats, or for a dedicated focus. Knowing what prompts to use in a given situation can help cut down the time you spend trying to create the right content with your tools.
  • Your content should be relevant and informative. Don’t just push out articles and blogs for the sake of it. If it isn’t relative and informative, it has little to no value for any reader. 
  • Make sure to add your own take on the content—add in your expertise, your thoughts, and your comments on the matter. This human touch not only displays your knowledge of the topic but also transforms the content into something that will be uniquely yours

What to Avoid When Using AI for Content Creation

And here’s what to avoid doing completely with your tools: 

  • Relying completely on AI for your content. As I said earlier, doing this will do you more harm than good, especially since the tools (and Google’s algorithm) can sniff out purely AI-written content. 
  • Substituting AI-generated content for human creativity. Use these tools to cut down on writing time so you and your team can invest in something more valuable—your human creativity, insight, and expertise. Always insert that into your writing.
  • Using AI for complicated, niche topics. Depending on your industry, you might not always be able to use it for your content. While AI is pretty sophisticated in what it can answer, it often cannot answer questions that require inference, a nuanced understanding of language, or a deeper understanding of multiple topics. 
  • Keyword stuffing. This is an SEO content best practice as old as time. When you stuff your writing full of keywords, it will make your work sound spammy, and may even get you penalized by Google for being spam content.
  • Forgetting to proofread and fact-check. Don’t expect AI to be perfect. Always proofread and fact-check. Any data, statistics, or facts given by your tool should be checked for accuracy. 

Conclusion

Can Google detect AI-generated content? The answer is yes, but that doesn’t mean that you have to stop using it in your SEO. 

Though several AI writing detectors are out there—many of which are free to use—it’s clear to me that AI-generated content won’t be going away anytime soon. In fact, there are several ways for it to be used ethically for SEO

With the proper guidelines, you can add AI writing tools as a valuable addition to your content creation arsenal, without fear of penalization.



Source link

Keep an eye on what we are doing
Be the first to get latest updates and exclusive content straight to your email inbox.
We promise not to spam you. You can unsubscribe at any time.
Invalid email address

SEO

Mozilla VPN Security Risks Discovered

Published

on

By

Mozilla VPN Security Risks Discovered

Mozilla published the results of a recent third-party security audit of its VPN services as part of it’s commitment to user privacy and security. The survey revealed security issues which were presented to Mozilla to be addressed with fixes to ensure user privacy and security.

Many search marketers use VPNs during the course of their business especially when using a Wi-Fi connection in order to protect sensitive data, so the  trustworthiness of a VNP is essential.

Mozilla VPN

A Virtual Private Network (VPN), is a service that hides (encrypts) a user’s Internet traffic so that no third party (like an ISP) can snoop and see what sites a user is visiting.

VPNs also add a layer of security from malicious activities such as session hijacking which can give an attacker full access to the websites a user is visiting.

There is a high expectation from users that the VPN will protect their privacy when they are browsing on the Internet.

Mozilla thus employs the services of a third party to conduct a security audit to make sure their VPN is thoroughly locked down.

Security Risks Discovered

The audit revealed vulnerabilities of medium or higher severity, ranging from Denial of Service (DoS). risks to keychain access leaks (related to encryption) and the lack of access controls.

Cure53, the third party security firm, discovered and addressed several risks. Among the issues were potential VPN leaks to the vulnerability of a rogue extension that disabled the VPN.

The scope of the audit encompassed the following products:

  • Mozilla VPN Qt6 App for macOS
  • Mozilla VPN Qt6 App for Linux
  • Mozilla VPN Qt6 App for Windows
  • Mozilla VPN Qt6 App for iOS
  • Mozilla VPN Qt6 App for Androi

These are the risks identified by the security audit:

  • FVP-03-003: DoS via serialized intent
  • FVP-03-008: Keychain access level leaks WG private key to iCloud
  • VP-03-010: VPN leak via captive portal detection
  • FVP-03-011: Lack of local TCP server access controls
  • FVP-03-012: Rogue extension can disable VPN using mozillavpnnp (High)

The rogue extension issue was rated as high severity. Each risk was subsequently addressed by Mozilla.

Mozilla presented the results of the security audit as part of their commitment to transparency and to maintain the trust and security of their users. Conducting a third party security audit is a best practice for a VPN provider that helps assure that the VPN is trustworthy and reliable.

Read Mozilla’s announcement:
Mozilla VPN Security Audit 2023

Featured Image by Shutterstock/Meilun

Source link

Keep an eye on what we are doing
Be the first to get latest updates and exclusive content straight to your email inbox.
We promise not to spam you. You can unsubscribe at any time.
Invalid email address
Continue Reading

SEO

Link Building Outreach for Noobs

Published

on

Link Building Outreach for Noobs

Link outreach is the process of contacting other websites to ask for a backlink to your website.

For example, here’s an outreach email we sent as part of a broken link building campaign:

In this guide, you’ll learn how to get started with link outreach and how to get better results. 

How to do link outreach

Link outreach is a four-step process:

1. Find prospects

No matter how amazing your email is, you won’t get responses if it’s not relevant to the person you’re contacting. This makes finding the right person to contact equally as important as crafting a great email.

Who to reach out to depends on your link building strategy. Here’s a table summarizing who you should find for the following link building tactics:

As a quick example, here’s how you would find sites likely to accept your guest posts:

  1. Go to Content Explorer
  2. Enter a related topic and change the dropdown to “In title”
  3. Filter for English results
  4. Filter for results with 500+ words
  5. Go to the “Websites” tab
Finding guest blogging opportunities via Content ExplorerFinding guest blogging opportunities via Content Explorer

This shows you the websites getting the most search traffic to content about your target topic.

From here, you’d want to look at the Authors column to prioritize sites with multiple authors, as this suggests that they may accept guest posts.

The Authors column indicate how many authors have written for the siteThe Authors column indicate how many authors have written for the site

If you want to learn how to find prospects for different link building tactics, I recommend reading the resource below.

2. Find their contact details

Once you’ve curated a list of people to reach out to, you’ll need to find their contact information.

Typically, this is their email address. The easiest way to find this is to use an email lookup tool like Hunter.io. All you need to do is enter the first name, last name, and domain of your target prospect. Hunter will find their email for you:

Finding Tim's email with Hunter.ioFinding Tim's email with Hunter.io

To prevent tearing your hair from searching for hundreds of emails one-by-one, most email lookup tools allow you to upload a CSV list of names and domains. Hunter also has a Google Sheets add-on to make this even easier.

Using the Hunter for Sheets add-on to find emails in bulk directly in Google SheetsUsing the Hunter for Sheets add-on to find emails in bulk directly in Google Sheets

3. Send a personalized pitch

Knowing who to reach out to is half the battle won. The next ‘battle’ to win is actually getting the person to care.

Think about it. For someone to link to you, the following things need to happen:

  • They must read your email
  • They must be convinced to check out your content
  • They must open the target page and complete all administrative tasks (log in to their CMS, find the link, etc.)
  • They must link to you or swap out links

That’s a lot of steps. Most people don’t care enough to do this. That’s why there’s more to link outreach than just writing the perfect email (I’ll cover this in the next section).

For now, let’s look at how to craft an amazing email. To do that, you need to answer three questions:

  1. Why should they open your email? — The subject line needs to capture attention in a busy inbox.
  2. Why should they read your email? — The body needs to be short and hook the reader in.
  3. Why should they link to you? — Your pitch needs to be compelling: What’s in it for them and why is your content link-worthy?

For example, here’s how we wrote our outreach email based on the three questions:

An analysis of our outreach email based on three questionsAn analysis of our outreach email based on three questions

Here’s another outreach email we wrote, this time for a campaign building links to our content marketing statistics post:

An analysis of our outreach email based on three questionsAn analysis of our outreach email based on three questions

4. Follow up, once

People are busy and their inboxes are crowded. They might have missed your email or read it and forgot.

Solve this by sending a short polite follow-up.

Example follow-up emailExample follow-up email

One is good enough. There’s no need to spam the other person with countless follow-up emails hoping for a different outcome. If they’re not interested, they’re not interested.

Link outreach tips

In theory, link outreach is simply finding the right person and asking them for a link. But there is more to it than that. I’ll explore some additional tips to help improve your outreach.

Don’t over-personalize

Some SEOs swear by the sniper approach to link outreach. That is: Each email is 100% customized to the person you are targeting.

But our experience taught us that over-personalization isn’t better. We ran link-building campaigns that sent hyper-personalized emails and got no results.

It makes logical sense: Most people just don’t do favors for strangers. I’m not saying it doesn’t happen—it does—but rarely will your amazing, hyper-personalized pitch change someone’s mind.

So, don’t spend all your time tweaking your email just to eke out minute gains.

Avoid common templates

My first reaction seeing this email is to delete it:

A bad outreach emailA bad outreach email

Why? Because it’s a template I’ve seen many times in my inbox. And so have many others.

Another reason: Not only did he reference a post I wrote six years ago, it was a guest post, i.e., I do not have control over the site. This shows why finding the right prospects is important. He even got my name wrong.

Templates do work, but bad ones don’t. You can’t expect to copy-paste one from a blog post and hope to achieve success.

A better approach is to use the scoped shotgun approach: use a template but with dynamic variables.

Email outreach template with dynamic variablesEmail outreach template with dynamic variables

You can do this with tools like Pitchbox and Buzzstream.

This can help achieve a decent level of personalization so your email isn’t spammy. But it doesn’t spend all your time writing customized emails for every prospect.

Send lots of emails

When we polled 800+ people on X and LinkedIn about their link outreach results, the average conversion rate was only 1-5%.

Link outreach conversion rates in 2023Link outreach conversion rates in 2023

This is why you need to send more emails. If you run the numbers, it just makes sense:

  • 100 outreach emails with a 1% success rate = 1 link
  • 1,000 outreach emails with a 1% success rate = 10 links

I’m not saying to spam everyone. But if you want more high-quality links, you need to reach out to more high-quality prospects.

Build a brand

A few years ago, we published a link building case study:

  • 515 outreach emails
  • 17.55% reply rate
  • 5.75% conversion rate

Pretty good results! Except the top comments were about how we only succeeded because of our brand:

Comments on our YouTube video saying we succeeded because of our brandComments on our YouTube video saying we succeeded because of our brand

It’s true; we acknowledge it. But I think the takeaway here isn’t that we should repeat the experiment with an unknown website. The takeaway is that more SEOs should be focused on building a brand.

We’re all humans—we rely on heuristics to make judgments. In this case, it’s branding. If your brand is recognizable, it solves the “stranger” problem—people know you, like you, and are more likely to link.

The question then: How do you build a brand?

I’d like to quote our Chief Marketing Officer Tim Soulo here:

What is a strong brand if not a consistent output of high-quality work that people enjoy? Ahrefs’ content team has been publishing top-notch content for quite a few years on our blog and YouTube channel. Slowly but surely, we were able to reach tens of millions of people and instill the idea that “Ahrefs’ content = quality content”—which now clearly works to our advantage.

Tim SouloTim Soulo

Ahrefs was once unknown, too. So, don’t be disheartened if no one is willing to link to you today. Rome wasn’t built in a day.

Trust the process and create incredible content. Show it to people. You’ll build your brand and reputation that way.

Build relationships with people in your industry

Outreach starts before you even ask for a link.

Think about it: People don’t do favors for strangers but they will for friends. If you want to build and maintain relationships in the industry, way before you start any link outreach campaigns.

Don’t just rely on emails either. Direct messages (DMs) on LinkedIn and X, phone calls—they all work. For example, Patrick Stox, our Product Advisor, used to have a list of contacts he regularly reached out to. He’d hop on calls and even send fruit baskets.

Create systems and automations

In its most fundamental form, link outreach is really about finding more people and sending more emails.

Doing this well is all about building systems and automations.

We have a few videos on how to build a team and a link-building system, so I recommend that you check them out.

Final thoughts

Good link outreach is indistinguishable from good business development.

In business development, your chances of success will increase if you:

  • Pitch the right partners
  • Have a strong brand
  • Have prior relationships with them
  • Pitch the right collaboration ideas

The same goes for link outreach. Follow the principles above and you will see more success for your link outreach campaigns.

Any questions or comments? Let me know on Twitter X.



Source link

Keep an eye on what we are doing
Be the first to get latest updates and exclusive content straight to your email inbox.
We promise not to spam you. You can unsubscribe at any time.
Invalid email address
Continue Reading

SEO

Research Shows Tree Of Thought Prompting Better Than Chain Of Thought

Published

on

By

Research Shows Tree Of Thought Prompting Better Than Chain Of Thought

Researchers discovered a way to defeat the safety guardrails in GPT4 and GPT4-Turbo, unlocking the ability to generate harmful and toxic content, essentially beating a large language model with another large language model.

The researchers discovered that the use of tree-of-thought (ToT)reasoning to repeat and refine a line of attack was useful for jailbreaking another large language model.

What they found is that the ToT approach was successful against GPT4, GPT4-Turbo, and PaLM-2, using a remarkably low number of queries to obtain a jailbreak, on average less than thirty queries.

Tree Of Thoughts Reasoning

A Google research paper from around May 2022 discovered Chain of Thought Prompting.

Chain of Thought (CoT) is a prompting strategy used on a generative AI to make it follow a sequence of steps in order to solve a problem and complete a task. The CoT method is often accompanied with examples to show the LLM how the steps work in a reasoning task.

So, rather than just ask a generative AI like Midjourney or ChatGPT to do a task, the chain of thought method instructs the AI how to follow a path of reasoning that’s composed of a series of steps.

Tree of Thoughts (ToT) reasoning, sometimes referred to as Tree of Thought (singular) is essentially a variation and improvement of CoT, but they’re two different things.

Tree of Thoughts reasoning is similar to CoT. The difference is that rather than training a generative AI to follow a single path of reasoning, ToT is built on a process that allows for multiple paths so that the AI can stop and self-assess then come up with alternate steps.

Tree of Thoughts reasoning was developed in May 2023 in a research paper titled Tree of Thoughts: Deliberate Problem Solving with Large Language Models (PDF)

The research paper describes Tree of Thought:

“…we introduce a new framework for language model inference, Tree of Thoughts (ToT), which generalizes over the popular Chain of Thought approach to prompting language models, and enables exploration over coherent units of text (thoughts) that serve as intermediate steps toward problem solving.

ToT allows LMs to perform deliberate decision making by considering multiple different reasoning paths and self-evaluating choices to decide the next course of action, as well as looking ahead or backtracking when necessary to make global choices.

Our experiments show that ToT significantly enhances language models’ problem-solving abilities…”

Tree Of Attacks With Pruning (TAP)

This new method of jailbreaking large language models is called Tree of Attacks with Pruning, TAP. TAP uses two LLMs, one for attacking and the other for evaluating.

TAP is able to outperform other jailbreaking methods by significant margins, only requiring black-box access to the LLM.

A black box, in computing, is where one can see what goes into an algorithm and what comes out. But what happens in the middle is unknown, thus it’s said to be in a black box.

Tree of thoughts (TAP) reasoning is used against a targeted LLM like GPT-4 to repetitively try different prompting, assess the results, then if necessary change course if that attempt is not promising.

This is called a process of iteration and pruning. Each prompting attempt is analyzed for the probability of success. If the path of attack is judged to be a dead end, the LLM will “prune” that path of attack and begin another and better series of prompting attacks.

This is why it’s called a “tree” in that rather than using a linear process of reasoning which is the hallmark of chain of thought (CoT) prompting, tree of thought prompting is non-linear because the reasoning process branches off to other areas of reasoning, much like a human might do.

The attacker issues a series of prompts, the evaluator evaluates the responses to those prompts and then makes a decision as to what the next path of attack will be by making a call as to whether the current path of attack is irrelevant or not, plus it also evaluates the results to determine the likely success of prompts that have not yet been tried.

What’s remarkable about this approach is that this process reduces the number of prompts needed to jailbreak GPT-4. Additionally, a greater number of jailbreaking prompts are discovered with TAP than with any other jailbreaking method.

The researchers observe:

“In this work, we present Tree of Attacks with Pruning (TAP), an automated method for generating jailbreaks that only requires black-box access to the target LLM.

TAP utilizes an LLM to iteratively refine candidate (attack) prompts using tree-of-thoughts reasoning until one of the generated prompts jailbreaks the target.

Crucially, before sending prompts to the target, TAP assesses them and prunes the ones unlikely to result in jailbreaks.

Using tree-of-thought reasoning allows TAP to navigate a large search space of prompts and pruning reduces the total number of queries sent to the target.

In empirical evaluations, we observe that TAP generates prompts that jailbreak state-of-the-art LLMs (including GPT4 and GPT4-Turbo) for more than 80% of the prompts using only a small number of queries. This significantly improves upon the previous state-of-the-art black-box method for generating jailbreaks.”

Tree Of Thought (ToT) Outperforms Chain Of Thought (CoT) Reasoning

Another interesting conclusion reached in the research paper is that, for this particular task, ToT reasoning outperforms CoT reasoning, even when adding pruning to the CoT method, where off topic prompting is pruned and discarded.

ToT Underperforms With GPT 3.5 Turbo

The researchers discovered that ChatGPT 3.5 Turbo didn’t perform well with CoT, revealing the limitations of GPT 3.5 Turbo. Actually, GPT 3.5 performed exceedingly poorly, dropping from 84% success rate to only a 4.2% success rate.

This is their observation about why GPT 3.5 underperforms:

“We observe that the choice of the evaluator can affect the performance of TAP: changing the attacker from GPT4 to GPT3.5-Turbo reduces the success rate from 84% to 4.2%.

The reason for the reduction in success rate is that GPT3.5-Turbo incorrectly determines that the target model is jailbroken (for the provided goal) and, hence, preemptively stops the method.

As a consequence, the variant sends significantly fewer queries than the original method…”

What This Mean For You

While it’s amusing that the researchers use the ToT method to beat an LLM with another LLM, it also highlights the usefulness of ToT for generating surprising new directions in prompting in order to achieve higher levels of output.

  • TL/DR Takeaways:
  • Tree of Thought prompting outperformed Chain of Thought methods
  • GPT 3.5 worked significantly poorly in comparison to GPT 4 in ToT
  • Pruning is a useful part of a prompting strategy
  • Research showed that ToT is superior to CoT in an intensive reasoning task like jailbreaking an LLM

Read the original research paper:

Tree of Attacks: Jailbreaking Black-Box LLMs Automatically (PDF)

Featured Image by Shutterstock/THE.STUDIO

Source link

Keep an eye on what we are doing
Be the first to get latest updates and exclusive content straight to your email inbox.
We promise not to spam you. You can unsubscribe at any time.
Invalid email address
Continue Reading

Trending