Best AI tools of 2023: ChatGPT rules the roost, followed by Google Bard, Midjourney, and more

ChatGPT

OpenAI unveiled its sensational chatbot ChatGPT in November 2022. The chatbot that could answer almost any question under the sky, ideate, and write programs stunned the world with its extraordinary capabilities. Chat Generative Pre Trained Transformer (ChatGPT) utilises OpenAI’s proprietary large language model (LLM), the GPT-series. After its launch, the San Francisco-based AI start-up went on to add several new features, launched mobile versions for iOS and Android and introduced GPT-4, its most potent AI model. 

While GPT-4 expanded the capabilities, OpenAI also endowed the chatbot with the power of the Internet and added plugins for subscribers to use ChatGPT for a variety of use cases. Towards the end of the year, OpenAI held its first-ever developer conference, opening its door to millions of developers. ChatGPT was enhanced with CustomGPTs, which allowed users to create their chatbots. Moreover, DALL-E 3 has been integrated into the interface of ChatGPT. Also, the newly announced GPT-4 Turbo comes with a context window of 1,28,000 tokens. The company has announced that it will soon be launching a GPT Store in the coming months. While the chatbot’s capabilities dominated headlines, November saw OpenAI marred by a boardroom coup that shook the company.

Bing AI Chat/Microsoft Copilot

OpenAI may require one to subscribe to its ChatGPT Plus to experience its mighty GPT-4. Bing AI Chat, introduced by Microsoft earlier this year, is a sophisticated feature of the Bing search engine. It is powered by GPT-4 and uses AI for conversational interactions. Backed by the advanced language models from OpenAI, Bing AI Chat can perform an assortment of tasks such as content creation, idea generation, and image generation. Bing AI Chat has been praised by users for its ability to engage in human-like conversations, assist users with coding, travel planning, learning languages, and more. Bing AI Chat which is now known as Microsoft Copilot is seen as a significant step towards integrating AI into search engines, offering an interactive and versatile search experience. 

Google Bard

Even as ChatGPT won over users around the world, in March 2023, Google introduced its conversational generative AI chatbot, Google Bard. It is currently available in 46 languages across 238 countries. Google Bard is the next leap in LLM technology and has evolved from Google’s proprietary LaMDA to PaLM and is at the moment running on Gemini. Google Bard can be called a search engine on steroids as it can offer informative conversations, translate many languages, be capable of creative writing, and even answer the most complex questions. Google Bard’s brain comprises a massive dataset of code and text which allows it to learn and understand human language. Its sophisticated algorithms enable it to process information, generate responses, and contextualise them. Essentially, Bard is a personal AI assistant that can learn and evolve with the user. Google says that it is still under development, however, it offers a glimpse into the future and our interactions with AI.

Runway Gen-2

Gen-2, developed by Runway, an innovative startup backed by Google, is a groundbreaking advancement in AI video software. This technology has already impressed the entertainment industry, as demonstrated by its use in the critically acclaimed Hollywood film ‘Everything Everywhere All at Once’, where it played a crucial role in creating stunning visual effects. Gen-2 is known for its ability to generate videos from various inputs. These inputs can range from simple text prompts to complex images and pre-existing videos. Our hands-on experience with Gen-2 revealed its impressive capabilities. While the tool is not without its limitations, the quality of the output is noteworthy. Areas for improvement include enhancing the precision of video generation, especially when working with detailed descriptions or specific image prompts. This level of refinement could further augment Gen-2’s functionality, making it an even more powerful tool for video creation and editing.

Dall-E 3

DALL-E is a generative AI model renowned for its image-generation capabilities. First launched in 2021, OpenAI introduced its third iteration earlier this year. This model generates images based on natural language inputs, commonly referred to as prompts. Currently, DALL-E 3 is accessible through ChatGPT, Microsoft Paint, Bing Image Creator, and other services using its API. Although these platforms offer similar experiences, ChatGPT is the official channel, making it the preferred method for using the AI model. According to an OpenAI website post, ‘DALL-E 3 is natively integrated with ChatGPT, enabling users to employ ChatGPT as a brainstorming partner and prompt refiner. Simply tell ChatGPT what you envision, ranging from a simple sentence to a detailed paragraph.’

OpenAI has also implemented measures to restrict DALL-E 3’s ability to generate violent, adult, or hateful content. To use DALL-E 3, users must subscribe to ChatGPT Plus, available for $20 per month or Rs 1999. It is also accessible to ChatGPT Enterprise users.

Midjourney

In simple terms, Midjourney acts like a magic paintbrush controlled by words. This wildly popular AI tool generates breathtaking images based on text descriptions and prompts. To use Midjourney, users should provide a detailed prompt describing the desired scene, style, ambiance, and mood. Specificity in colour, texture, composition, and lighting settings is essential. Midjourney analyses these prompts and employs powerful algorithms that blend millions of images and artistic styles to match the user’s request. Within seconds, the AI tool produces unique artwork, which can then be refined, tweaked, or used to generate more variations. This year, images of Pope Francis in a Balenciaga puffer jacket, created using Midjourney, went viral on the internet. Many initially believed these images to be real, showcasing Midjourney’s ability to create precise and photorealistic images.

Pi Chatbot

In August, Mustafa Suleyman, a former co-founder of DeepMind, and Reid Hoffman, an ex-co-founder of LinkedIn, launched an AI chatbot through their new startup, Inflection AI. This unique chatbot, named Pi, offers a fresh perspective on AI personal assistants. Pi differs significantly from ChatGPT and Google Bard. Its creators describe it as a chatbot designed to be ‘supportive, smart, and always there for you.’ Unlike typical chatbots that primarily handle mundane tasks, Pi is intended to be more of a companion to humans.

The company explained in a tweet, ‘Pi stands for “personal intelligence” and provides infinite knowledge tailored to your unique needs.’ We used Pi as motivation to overcome anxiety about driving on the highway and were impressed by its heightened empathy and human-like conversational abilities. The web version of the chatbot provides a pleasant experience, featuring a dreamy layout and interface where text floats invitingly, responding with empathetic emojis and words. Now, the chatbot also has real-time access to latest information from the web. 

Ever wanted to ask Pi about the latest news, events, and more? Now you can because it's fully up to date with real-time access to fresh information from across the Web.

What are you going to talk to Pi about today? pic.twitter.com/tIdmIR3sIm

— Inflection AI (@inflectionAI) October 16, 2023

Claude 2

Anthropic is the San Francisco-based AI startup that made headlines this year after it unveiled its Claude 2 chatbot in July. Claude 2 was a commercial version of its Claude chatbot that was earlier limited to closed beta in Slack. Claude 2 is an AI-powered conversational agent that has been designed to be helpful, harmless, and honest, through a concept known as Constitutional AI. 

Claude was the first chatbot to come with a big context window of 1,00,000 tokens which equals 75,000 words. This essentially meant that one could use an entire novel as a prompt. At that point in time, ChatGPT was limited to 8,912 tokens. Eventually, OpenAI surpassed the token window with GPT-4 Turbo. Besides, Claude 2 was mainly used to have natural conversations and assist with various tasks like answering questions, making recommendations, and more. The chatbot is enabled with self-supervision learning which allows it to train from public dialogue instead of manual labelling of data. It is based on transformer-based neural networks that enable it to understand and generate human-like text. 

Character.ai

Character AI is a chatbot that lets users have conversations with celebrities, historical figures, and fictional characters. It’s different from ChatGPT and Google Bard as its conversations seem more natural. The tool was created with natural language processing and machine learning, which makes its responses more engaging and relevant. In August, it was reported that Character AI had visitors spending the most time on it, an average eight times higher than ChatGPT. The chatbot features AI versions of a famous footballer, renowned physicist, celebrated pop star, and others. Additionally, users can engage in conversations with language educators, mental health counsellors, and well-known characters from video games.

GitHub Copilot

GitHub Copilot is essentially an AI pair programmer designed to enhance the creativity involved in coding. The tool developed by GitHub, which is owned by Microsoft is a collaborative tool that provides contextual suggestions based on comments and code along with offering real-time assistance to developers. GitHub Copilot is capable of suggesting not only individual lines but entire functions almost instantly, and this streamlines the coding process significantly. It is available as an extension in popular Integrated Development Environments (IDE) such as Visual Studio Code, NeoVim, and JetBrains suite of IDEs. Copilot also assists in addressing several coding-related problems and is capable of keeping developers in the flow. It supports all major programming languages and has the intelligence to adapt to a user’s style of coding over time. 

Adobe Firefly 

This year when it comes to creativity, Adobe’s Firefly, accessible at adobe.com, has been one of the go-to AI tools for creators this year. This AI image generator is not justany tool, it’s a creative powerhouse as it lets anyone add colour to their ideas. Imagine a scenario where you have an idea, a scene, or a concept, one simply needs to input the same into Firefly and let it do the work. It almost instantly transforms your prompts into stunning high-quality images. This tool is perfect for content creators, graphic designers, and almost anyone who needs visual content. In our experience, with Firefly, you can effortlessly create unique thumbnails for videos, striking graphics for social media, or even artwork for your blogs and websites. Firefly is currently free as it is in its beta phase.

Perplexity AI

Ever since the launch of ChatGPT, one thing that was constantly discussed was embedding AI into conventional search engines. Perplexity AI by San Francisco software development company is a conversational AI search engine that offers a chatbot-like interface allowing users to ask questions in natural language. In response, the chatbot responds by citing sources from around the web. It also allows users to ask follow-up questions to dive deeper into any topic. In our experience, Perplexity AI is that bridge that connects creativity and knowledge. The search engine functions like a virtual librarian such as offering precise answers with a source to any query. Users of Perplexity AI have hailed it as an invaluable tool for research, academic purposes, content creation, and even enhancing one’s knowledge. Perplexity AI focuses on not just delivering information but also emphasises on quality and credibility of the content. 

Reflecting on 2023: A transformative year where we redefined the future of search. Thanks for being part of our journey. The team is excited to serve you all, by providing accurate, fast answers to all your questions! Here's to more exploration and knowledge in 2024! pic.twitter.com/hOIYy0RaBn

— Perplexity (@perplexity_ai) December 23, 2023

What lies ahead

The tools assembled here have been listed based on factors such as innovation, utility, and user base. These tools and their creators have also raised some pertinent questions about the future of AI assistants and co-pilots.

2024 is drawing closer. The year 2023 can be called a transformative year for AI with some groundbreaking advancements and trends. We anticipate that the rapid pace of innovations in AI will continue next year as well. Domains such as natural language processing, enhanced AI ethics protocols, and more intuitive user interfaces are likely to dominate in 2024. We also feel that concerns such as those surrounding data privacy, algorithmic bias, and the impact of AI on jobs are likely to persist. We also expect to see the integration of AI in sustainable development, healthcare, and education. However, with these developments, there will be a need for greater monitoring to ensure that AI is used ethically and there is equitable access.