ChatGPT vs. HuggingChat: Which Is Better?

BY

BOB SHARP

PUBLISHED MAY 12, 2023

HuggingChat is gaining many admirers, but can it really compete with ChatGPT?

Readers like you help support MUO. When you make a purchase using links on our site, we may earn an affiliate commission. Read More.

When ChatGPT burst spectacularly onto the scene, it was inevitable that similar tools would soon follow. One of these is HuggingChat. At first glance, HuggingChat and ChatGPT look and behave very similarly. Both tools perform similar functions, have similar interfaces, and use AI to produce responses.

However, dig a little deeper, and notable differences become quickly apparent.

Close

922.8K

EcoFlow Glacier: Love Partying? You HAVE to Check Out This New Cooler

What Is HuggingChat?

We won't go into too much detail here, as by now, most people know what generative AI chatbots are. But it is worth touching on, as this will help you understand some underlying differences between the two chatbots.

Put simply, HuggingChat is an open-source alternative to ChatGPT. Open-source means that the code is freely available to the public. This gives the platform customization options that platforms like ChatGPT lack.

Broadly speaking, HuggingChat can perform the same functions as its closed-source cousin, ChatGPT:

• Writing assistance: Writing tips, ideas, grammar corrections, email, and article composition.

• Technical Support: An embedded technical knowledge base allows the tool to be used as a technical assistant.

• Natural Language Search: HuggingChat allows you to search the internet without relying on keyword queries.

Now that we know all about HuggingChat, let's see how it stacks up against ChatGPT.

HuggingChat vs. ChatGPT: How Do They Compare?

Let's start by comparing how to access the tools. ChatGPT first, although it is possible to use ChatGPT without an account, accessing the tool directly will require an OpenAI account. Once you have set this up, the tool is free to use (unless you upgrade to ChatGPT Pro).

Now, with HuggingChat, the waters are a bit murkier. If asked, HuggingChat will assure you that an account is required, as the image below confirms.

However, we used it with an account and without any problems. If this changes, or you'd prefer to set up an account, it is free, and you can sign up on HuggingFace.

The tools have remarkably similar interfaces, as the bulk of these tools do. Simplicity is the key to AI chatbots, and both tools have a simple input box to allow you to enter prompts, which is about it. However, learning prompting techniques can help improve response quality.

Next, we will look at other metrics and capabilities and compare the responses.

Large Language Models (LLMs)

The beating heart of any AI chatbot is the LLM. These are the underlying datasets used to generate human-like responses to prompts. Each tool uses a different LLM, as described below.

• ChatGPT: ChatGPT is based on OpenAI's GPT family of LLMs. More specifically, the free version of the tool uses GPT-3.5, and Premium users have access to GPT-4. ChatGPT is restricted to using data collected up to late 2021 only.

• HuggingChat: HuggingChat is based on Meta's LlaMA LLM. This has the advantage of access to data collected up to April 12, 2023.

This should give HuggingChat an advantage when searching for current data. We tested this by asking both tools the same question: Who won the Baseball World Series in 2022?

As you can see, HuggingChat answered this easily.

Whereas, ChatGPT appears to believe that 2022 lies in the future!

Summarizing Articles

You can use URLs to "force" both tools to access more current data. We tested this ability by asking them to summarize a MakeUseOf article on remotely accessing Android phones. In the first instance, we provided the URL.

Both tools managed this with the ChatGPT response edging it in quality as it summarized the tools mentioned in the article.

The HuggingChat response wasn't bad but lacked some of the detail covered by ChatGPT.

To test this ability further, we asked a similar question without entering the URL and using an article that was only published (it was "What Are IPFS Phishing Attacks," for reference). Again, both tools managed this admirably, as the images demonstrate.

This was ChatGPT's effort:

HuggingChat's summary looked like this:

There isn't much difference between the two tools here, so let's see how they compare when we test them for creativity.

HuggingChat vs. ChatGPT: How Do They Compare Creatively?

Although surrounded by controversy, the ability to create creative works is perhaps one of the most useful aspects of AI chatbots. However, we will sidestep the moral and ethical implications and put the tools head-to-head in a creative clash of the chatbots.

Creativity is a difficult metric to quantify. What is poetry to your ears may be unintelligible nonsense to the next person. So, all we will do here is set an identical task for both tools and let you decide which reply buzzed your buttons.

We asked both to compose a two-verse love song based on Romeo and Juliet. This was ChatGPT's response:

And this was HuggingChat's.

Shakespeare himself would be impressed with both of these!

Security and Privacy

Security and privacy are general concerns surrounding generative AI. There are definite security concerns with ChatGPT, but it is fair to say that most of these concerns apply to most chatbots, including HuggingChat.

The problem is that this fledgling technology has plenty of rough edges. Security is certainly one of these. These tools are incredibly powerful, but care is needed as security and privacy issues include:

• Data privacy: The training data contains personal information and have access to user data.

• Malicious use: Bad actors can use it to generate spam and fake news, and it can also be used to aid in phishing attacks.

• Cybersecurity vulnerabilities: ChatGPT is a complex system that may contain cybersecurity vulnerabilities that can be exploited by attackers.

One thing to note is that HuggingChat seems to work fine with an account. This is useful if you want to minimize the privacy risks associated with the platforms.

Which Is Best: HuggingChat or ChatGPT?

As you can see from the tests, there is no easy answer. ChatGPT had a definite edge when summarizing articles, whereas HuggingChat had the advantage when you compare how current the training data is.

One of the big problems when trying to decide between the tools is that the results of any given prompt are unpredictable. In essence, this means that if you tweak a prompt slightly, the quality and accuracy of the results can vary wildly.

On the point of accuracy, both these tools acknowledge that the accuracy of the results may be flawed. You should always take this into account when using either of them.

We also found both tools to be responsive and quick with answers. This would seem to illustrate that ChatGPT has addressed the demand issues that slowed the platform down.

Ultimately, the choice between them will likely be driven more by user requirements than user preferences. As the tools are free and quick to access, checking both out is not a great hassle.

Are You a Hugger or a Chatter?

These tools represent the future of human and machine interaction. Both are incredibly powerful, and it wasn't so long ago that the thought of having such tools at your fingertips would have seemed inconceivable.

As AI tools continue to advance, the choice between HuggingChat and ChatGPT will ultimately depend on the user's specific needs. However, the unpredictable nature of their results highlights the importance of careful consideration and evaluation of each tool's strengths and weaknesses.

What these show us is that AI has reached a point of critical momentum. Tools such as HuggingChat and ChatGPT will continue to develop and improve, and no doubt, many others will follow.

ABOUT THE AUTHOR

Bob Sharp • Contributing Writer For Technology Explained

(34 Articles Published)

Bob is a professional writer who has been writing extensively on technology topics since 2020, with a focus on AI, software, hardware, SaaS, and networks. Prior to becoming a writer, he owned and operated an IT firm in Scotland for over twenty years.
He currently resides in Spain and continues to provide valuable insights and analysis on the latest developments in the tech industry.

ChatGPT vs. Bing Chat: What's the Best Generative AI Chatbot?

BY

BOB SHARP

PUBLISHED MAR 20, 2023

ChatGPT is making waves, but Microsoft's Bing Chat is also impressive. What should you use?

Image Credit: Koshiro K/Shutterstock

Readers like you help support MUO. When you make a purchase using links on our site, we may earn an affiliate commission. Read More.

Bing Chat and ChatGPT are the two public faces of artificial intelligence chatbots. Although the generative AI tools are similar and both use OpenAI's GPT AI model, distinct differences exist between the two.

As more platforms adopt GPT, the diversity of the technology will widen. So, looking at the early adopters and identifying how they're already differentiating is interesting.

So, what's the difference between Microsoft's Bing Chat and OpenAI's ChatGPT, and which generative AI chatbot should you use?

How to Use Bing Chat and ChatGPT

The first difference between Bing Chat and ChatGPT is in how you access each tool.

To access Bing Chat, you need the latest version of Microsoft Edge and have it logged onto a Microsoft account. Your computer should automatically update Edge to the latest version, but there are some ways you can force Windows to update if it doesn't.

Once this is done, you will see a prompt similar to the one below. The Bing Chat waiting list is now discontinued, but you may still see a prompt. The next step is to press the Chat now button, and that's it!

The process with ChatGPT is even simpler. First, you will need to create an OpenAI account. Once created, you can use ChatGPT from any supported browser. This multi-browser compatibility marks the first notable difference. Bing Chat does offer a form of multi-browser support but with limited functionality.

Comparing the Language Models of Bing Chat and ChatGPT

One of the fundamental differences between the chatbots is the language model they use. Currently, free ChatGPT users are restricted to GPT-3.5, with the much-hyped GPT-4 only available to premium users.

Bing Chat uses the latest version of the language model, GPT-4. However, there are some key differences between GPT-4 and GPT-3.5:

Creativity

Both have levels of creativity that are testimony to the heights that AI has reached. However, creativity is a difficult metric to measure, and the tests we tried showed GPT-3.5 is close to GPT-4 in simple creative tasks.

For example, below is Bing Chat's effort at a simple limerick.

As you see from the ChatGPT effort below, both models can construct creative prose remarkably well.

Safety

Moderating responses that were factually incorrect or otherwise unsuitable was performed "on the fly" with GPT-3.5. In other words, it was a reactive strategy that acted after the horse had bolted. GPT-4 has safety measures designed into the model, meaning safety is more proactive. Where safety is a concern, Bing Chat technically has the upper hand.

Image Processing and Accuracy

The ability to process image data is also a new feature in GPT-4. However, this feature currently isn't integrated into Bing Chat, so it isn't covered here. The other major difference is the accuracy of the response. This is covered next.

How Accurate Are Bing Chat and ChatGPT

Both platforms go to pains to make users aware that the models may generate responses that contain incorrect information. This is clearly displayed on both interfaces.

Bing Chat has a distinct advantage here. The GPT-4 model has the twin advantages of access to more recent data and multiple sources of information. By contrast, the GPT-3.5/GPT-4 dataset was cut off at the end of 2021, and though it has received some topical updates, it is limited.

Unlike testing the difference in creativity, testing this was easy. The results were comprehensive, with Bing Chat returning the most impressive answer to a straightforward question, "How many tons of plastic were recycled in 2020?"

The response supplied figures as well as links to the sources from which the data was retrieved. However, the inclusion of US plastic exports was slightly out of context with the question. We would hesitate to call this a glitch, but it does show AI's tendency to wander off-topic at times.

By contrast, the ChatGPT response was to admit its limitations.

The answer tries to provide estimates and other sources that could be referenced, but there are no hard and fast facts.

Comparing the Interfaces of ChatGPT and Bing Chat

There are similarities across both interfaces, which reflects their shared pedigree. For example, both have a simple search box and offer the option to regenerate a response after providing an answer.

However, ChatGPT adheres more strictly to the traditional sense of chatbots by not providing prompts that encourage follow-up questions. In contrast, Bing Chat offers recommendations for further information with each search result and options to obtain more creative, balanced, or precise chat responses.

Another significant difference is that scrolling down past the "Chat" section of the Bing website takes you to the Bing search engine, which also uses GPT-4. This makes Bing Chat an excellent choice for hybrid searching, where access to search engines and chatbots is necessary.

For instance, if you're searching for a list of online resources, Bing Chat provides links to the listed websites, while ChatGPT compiles a similar list of results without links. Furthermore, in Bing Chat, scrolling down to the search section provides a link to pursue your search further.

As you can see, the response included links to the listed websites. Compare this to the ChatGPT response:

Which Is Best: Bing Chat or Chat GPT?

There is no straightforward answer, as both tools are incredibly impressive.

A good analogy is to compare ChatGPT to a blank canvas, with Bing Chat being compared to paint with numbers. Bing Chat will take a more guiding role to get you to where you want. On the other hand, ChatGPT leaves more to your imagination and relies on user input to reach a desired result.

In these terms, both perform superbly. However, there are plus and negative sides to both platforms that can help you decide.

Bing Chat is the best option for those looking for up-to-date information and to assist with follow-on responses. Thanks to its more proactive approach, it is also a better option if online safety is a concern. However, there are plenty of ways that kids can safely use ChatGPT.

The downside is the need to use the latest version of Edge. There is also a 15-chat limit per session with a 2000-character limit on searches, which may be restrictive to some users.

Despite ChatGPT using GPT-3.5, we still found it better for tasks like creative writing. GPT's blank canvas approach and heavier focus on "chatting" were more useful.

Additionally, the ability to run on browsers other than Edge may make it more appealing to some.

Ultimately, this is a case of choosing the right tool for the job. The beauty of this is that both platforms are free, so trying them both is quick and easy.

Chatbots Are Here to Stay

As AI chatbots become more widespread, the differences between these early adopters will only continue to widen. However, by understanding the unique features of each platform, users can choose the chatbot that best suits their needs.

nsarrazin very basic rate limiter (#320)

922b1b23 days ago

Git stats

•  297 commits

Files

Type

Name

Latest commit message

Commit time

.github/workflows

✅ Add tests setup + test sign in flow

last month

.vscode

migrate to eslint-plugin-svelte from deprecated plugin (#182)

last month

src

very basic rate limiter (#320)

3 days ago

static

Added ChatUI branding & put theming behind an env variable (#298)

last week

.dockerignore

Add incremental build + multi layer docker for size reduction (#259)

last month

.env

very basic rate limiter (#320)

3 days ago

.eslintignore

init

4 months ago

.eslintrc.cjs

error on unused vars instead of warn

last month

.gitignore

init

4 months ago

.npmrc

init

4 months ago

.prettierignore

init

4 months ago

.prettierrc

Add tailwind prettier (#123)

2 months ago

Dockerfile

Add incremental build + multi layer docker for size reduction (#259)

last month

LICENSE

Create LICENSE (#194)

last month

PRIVACY.md

Bump to v0.2 (#229)

last month

README.md

add details about websearch to README

last week

package-lock.json

Fixed grammar (#291)

2 weeks ago

package.json

Bump version to v0.3 (#283)

3 weeks ago

postcss.config.js

🚨 Run prettier

2 months ago

svelte.config.js

Added ChatUI branding & put theming behind an env variable (#298)

last week

tailwind.config.cjs

Added ChatUI branding & put theming behind an env variable (#298)

last week

tsconfig.json

Polyfill safari 14 (#149)

last month

vite.config.ts

Prettier update (#54)

2 months ago

README.md

title

emoji

colorFrom

colorTo

sdk

pinned

license

base_path

app_port

chat-ui

🔥

purple

purple

docker

false

apache-2.0

/chat

3000

      Chat UI

A chat interface using open source models, eg OpenAssistant. It is a SvelteKit app and it powers the HuggingChat app on hf.co/chat.

1. Setup

2. Launch

3. Extra parameters

4. Deploying to a HF Space

5. Building

    Setup

The default config for Chat UI is stored in the .env file. You will need to override some values to get Chat UI to run locally. This is done in .env.local.

Start by creating a .env.local file in the root of the repository. The bare minimum config you need to get Chat UI to run locally is the following:

MONGODB_URL=<the URL to your mongoDB instance>

HF_ACCESS_TOKEN=<your access token>

    Database

The chat history is stored in a MongoDB instance, and having a DB instance available is needed for Chat UI to work.

You can use a local MongoDB instance. The easiest way is to spin one up using docker:

docker run -d -p 27017:27017 --name mongo-chatui mongo:latest

In which case the url of your DB will be MONGODB_URL=mongodb://localhost:27017.

Alternatively, you can use a free MongoDB Atlas instance for this, Chat UI should fit comfortably within the free tier. After which you can set the MONGODB_URL variable in .env.local to match your instance.

    Hugging Face Access Token

You will need a Hugging Face access token to run Chat UI locally, using the remote inference endpoints. You can get one from your Hugging Face profile.

   Launch

After you're done with the .env.local file you can run Chat UI locally with:

npm install

npm run dev

    Extra parameters

    OpenID connect

The login feature is disabled by default and users are attributed a unique ID based on their browser. But if you want to use OpenID to authenticate your users, you can add the following to your .env.local file:

OPENID_PROVIDER_URL=<your OIDC issuer>

OPENID_CLIENT_ID=<your OIDC client ID>

OPENID_CLIENT_SECRET=<your OIDC client secret>

These variables will enable the openID sign-in modal for users.

    Theming

You can use a few environment variables to customize the look and feel of chat-ui. These are by default:

PUBLIC_APP_NAME=ChatUI

PUBLIC_APP_ASSETS=chatui

PUBLIC_APP_COLOR=blue

PUBLIC_APP_DATA_SHARING=

PUBLIC_APP_DISCLAIMER=

• PUBLIC_APP_NAME The name used as a title throughout the app.

• PUBLIC_APP_ASSETS Is used to find logos & favicons in static/$PUBLIC_APP_ASSETS, current options are chatui and huggingchat.

• PUBLIC_APP_COLOR Can be any of the tailwind colors.

• PUBLIC_APP_DATA_SHARING Can be set to 1 to add a toggle in the user settings that lets your users opt-in to data sharing with models creator.

• PUBLIC_APP_DISCLAIMER If set to 1, we show a disclaimer about generated outputs on login.

    Web Search

You can enable the web search by adding either SERPER_API_KEY (serper.dev) or SERPAPI_KEY (serpapi.com) to your .env.local.

    Custom models

You can customize the parameters passed to the model or even use a new model by updating the MODELS variable in your .env.local. The default one can be found in .env and looks like this :

MODELS=`[

  {

    "name": "OpenAssistant/oasst-sft-4-pythia-12b-epoch-3.5",

    "datasetName": "OpenAssistant/oasst1",

    "description": "A good alternative to ChatGPT",

    "websiteUrl": "https://open-assistant.io",

    "userMessageToken": "<|prompter|>",

    "assistantMessageToken": "<|assistant|>",

    "messageEndToken": "</s>",

    "preprompt": "Below are a series of dialogues between various people and an AI assistant. The AI tries to be helpful, polite, honest, sophisticated, emotionally aware, and humble-but-knowledgeable. The assistant is happy to help with almost anything, and will do its best to understand exactly what is needed. It also tries to avoid giving false or misleading information, and it caveats when it isn't entirely sure about the right answer. That said, the assistant is practical and really does its best, and doesn't let caution get too much in the way of being useful.\n-----\n",

    "promptExamples": [

      {

        "title": "Write an email from bullet list",

        "prompt": "As a restaurant owner, write a professional email to the supplier to get these products every week: \n\n- Wine (x10)\n- Eggs (x24)\n- Bread (x12)"

      }, {

        "title": "Code a snake game",

        "prompt": "Code a basic snake game in python, give explanations for each step."

      }, {

        "title": "Assist in a task",

        "prompt": "How do I make a delicious lemon cheesecake?"

      }

    ],

    "parameters": {

      "temperature": 0.9,

      "top_p": 0.95,

      "repetition_penalty": 1.2,

      "top_k": 50,

      "truncate": 1000,

      "max_new_tokens": 1024

    }

  }

]`

You can change things like the parameters, or customize the preprompt to better suit your needs. You can also add more models by adding more objects to the array, with different preprompts for example.

      Running your own models using a custom endpoint

If you want to, you can even run your own models locally, by having a look at our endpoint project, text-generation-inference. You can then add your own endpoints to the MODELS variable in .env.local, by adding an "endpoints" key for each model in MODELS.

{

// rest of the model config here

"endpoints": [{"url": "https://HOST:PORT/generate_stream"}]

}

If endpoints is left unspecified, ChatUI will look for the model on the hosted Hugging Face inference API using the model name.

      Custom endpoint authorization

Custom endpoints may require authorization, depending on how you configure them. Authentication will usually be set either with Basic or Bearer.

For Basic we will need to generate a base64 encoding of the username and password.

echo -n "USER:PASS" | base64

VVNFUjpQQVNT

For Bearer you can use a token, which can be grabbed from here.

You can then add the generated information and the authorization parameter to your .env.local.

"endpoints": [

{

"url": "https://HOST:PORT/generate_stream",

"authorization": "Basic VVNFUjpQQVNT",

}

]

     Models hosted on multiple custom endpoints

If the model being hosted will be available on multiple servers/instances add the weight parameter to your .env.local. The weight will be used to determine the probability of requesting a particular endpoint.

"endpoints": [

{

"url": "https://HOST:PORT/generate_stream",

"weight": 1

}

{

"url": "https://HOST:PORT/generate_stream",

"weight": 2

}

...

]

    Deploying to a HF Space

Create a DOTENV_LOCAL secret to your HF space with the content of your .env.local, and they will be picked up automatically when you run.

    Building

To create a production version of your app:

npm run build

You can preview the production build with npm run preview.

To deploy your app, you may need to install an adapter for your target environment.