From novelty to mainstream productivity tool, AI has made quite the leap in 2023 and has integrated itself within the fabric of our day-to-day lives. We will certainly look back on 2023 as the year of the Generative AI boom.
Hundreds if not thousands of AI-related releases and announcements were made throughout the entire year, as the world followed the growing influence and transformative potential of AI across various sectors.
As we gear up to face the challenges and opportunities of 2024, let’s reflect on the most noteworthy AI advancements of 2023, and consider how they will continue to shape our future.
AI in 2023: The Timeline
January
🚀 AI Development
2023 kicked off with exciting news, as Microsoft announced a 10 billion dollars investment into OpenAI. Given OpenAI’s prominent position at the forefront of AI development, this wasn’t entirely a surprise. After all, ChatGPT garnered over 1 million users within just five days of its November 2022 release.
The company, also known for its text-to-image Generative AI model, Dall-E, was now poised to push the boundaries further in advancing its cutting-edge Generative AI technologies.
The same month, AI company ElevenLabs also made waves by publicly releasing its beta platform, and setting a new standard with the most sophisticated and lifelike text-to-speech software the world had ever seen.
Following this milestone, ElevenLabs successfully secured a $19 million Series A funding round, valuing the company at approximately $100 million, despite having only 15 employees and no office space at the time.
The downside of this groundbreaking AI technology, however, was the surge in voice scams, in response to which ElevenLabs introduced a voluntary detection tool, in an attempt to establish safeguards against malicious activities.
📷 AI Image Generation
January also brought forth breakthroughs in the world of AI-generated art. Researchers At UC Berkeley release Instruct pix2pix, an open source diffusion model based on Stable Diffusion that made AI-driven image editing possible.
Prior to this breakthrough, Generative AI models such as Dall-E, Stable Diffusion, and MidJourney were able to generate images based on text prompts, but editing images with text prompts was not possible. Hence, the introduction of Instruct pix2pix marked a pivotal leap forward in the area of AI imagery.
🚑 AI in Healthcare
The onset of 2023 also thrust the healthcare AI model Sybil into the public spotlight. Developed by researchers at MIT’s Abdul Latif Jameel Clinic for Machine Learning in Health, Mass General Cancer Center (MGCC), and Chang Gung Memorial Hospital (CGMH), Sybil emerged as an innovative lung cancer risk assessment tool.
AI model Sybil analyzes image data to forecast a patient’s likelihood of developing lung cancer within six years, taking computer tomography screening to new heights.
February
🚀 AI Development
In response to OpenAI’s ChatGPT, Google announced the release of Google Bard in February. Unfortunately, Google’s chatbot was off to a shameful start, sharing inaccurate information in its dedicated promotional ad. This misstep ended up tarnishing parent company Alphabet’s reputation post-release, resulting in a significant drop in share values.
But the chatbot rivalry was far from over. Only one day after Bard’s release, Microsoft unveiled AI-powered Microsoft Bing and Edge, “your copilot for the web”, with the goal being to leverage Open AI’s GPT models to reinvent search. However, Bing chat encountered its own challenges, prompting Microsoft to soon impose a limit of 50 questions per day after unsettling interactions with users.
📷 AI Image Generation
In February, the AI art world experienced a paradigm shift with the groundbreaking debut of ControlNet, a neural network capable of adding extra conditions to image generation in Stable Diffusion, such as specifying certain poses, replicating the composition of other images, or transforming scribbles into realistic images.
🤝 AI Partnerships
February also marked the announcement of a partnership between Hugging Face and AWS, aimed at enhancing the accessibility to AI.
Hugging Face outlined its strategy to utilize AWS as a cloud provider, enabling developers within their community to utilize AWS SageMaker and other AWS-powered tools for training, fine-tuning, and deploying machine learning models.
March
🚀 AI Development
One of the pivotal announcements in March came from OpenAI with the launch of GPT-4, accompanied by the rollout of ChatGPT plugins later in the month.
March 30th also saw the debut of Auto-GPT, an experimental, open-source Python application leveraging GPT-4 to execute tasks with minimal human intervention, effectively self-prompting until achieving its designated objective.
📷 AI Image & 📹 Video Generation
March proved to be a significant month for the AI art world as well, with the several noteworthy releases hitting the spotlight:
- The unveiling of MidJourney v5 showcased remarkable enhancements in image quality, stylistic range, and realism.
- Adobe introduced Firefly, empowering all Adobe Creative Cloud users to produce high-quality images and stunning effects effortlessly.
- Runway launched Gen-1, a tool enabling creators to generate new videos from existing ones based on text or image prompts.
✏️ AI-Powered Content Creation
In March, marketers also experienced a wave of AI-driven innovation, with Hubspot introducing ChatSpot.ai, alongside several other AI-powered content assistant tools designed to enhance productivity.
🎓 AI in Education
An interesting update came from well-known online learning platform Khan Academy, who launched the Khanmigo AI, an AI-powered tutor or teaching assistant. This development underscored the potential of AI in revolutionizing education and improving learning experiences.
🚗 AI in Automotive
In March, Ford also made the headlines with the announcement of its new subsidiary company, Latitude AI, tasked with developing new automated driving technology with an initial focus on a hands-free, eyes-off driver assist system for next-generation Ford vehicles.
April
🚀 AI Development
In April Meta released Segment Anything Model. As Meta explained, SAM would allow users to “segment objects with just a click or by interactively clicking points to include and exclude from the object”.
📹 AI Video Generation
April was also the month when startup WonderDynamics released AI tool WonderStudio, a platform that automatically animates and inserts computer generated characters into live-action scenes. The company was recognized by 80 Level as one of 2023’s most impactful tech innovations.
May
📰 AI News
The beginning of May brought with itself unexpected news, as ‘the godfather of AI' Geoffrey Hinton quit Google in order to be able to talk more freely about the perils of this advanced technology.
In the meantime, on May 10th Google announced their plan to “make AI helpful to everyone” by incorporating AI into Google Workspace.
📜 AI Regulation
May was also the month when Sam Altman, CEO of OpenAI testified before the Senate about the need to regulate the increasingly powerful AI technology created by his company and others such as Google and Microsoft.
At the hearing, Altman was joined by Christina Montgomery, IBM’s Chief Privacy and Trust Officer, and Gary Marcus, professor emeritus of psychology and neural science at New York University.
June
📷 AI Image & 📹 Video Generation
In June, Runway officially released Gen-2, a multimodal AI system able to generate new videos from text, image or video prompts. With significant enhancements over Gen-1, Gen-2 established a new benchmark for AI video generation.
The same month, Adobe revealed the integration of Firefly with Photoshop, sparking excitement among designers, marketers, and content creators worldwide. The new generative AI-driven fill tool enabled users to seamlessly alter images based on text prompts.
July
🚀 AI Development
Following a relatively slower month in June, July surged ahead with a flurry of new announcements:
- OpenAI introduced Code Interpreter, a tool enabling the Assistants API to write and execute Python code within a sandboxed environment.
- Anthropic, a key competitor of OpenAI, unveiled Claude 2, an AI chatbot trained to engage in natural, text-based conversations.
- Meta and Microsoft introduced the next-generation of Llama, with Llama 2 now available for both research and commercial use free of charge.
- OpenAI also launched Custom Instructions for ChatGPT, granting Plus users enhanced control over the chatbot’s responses.
August
📷 AI Image & 📹 Video Generation
In August, Midjourney rolled out its Vary Region Editor, empowering users to meticulously select and regenerate specific portions of an upscaled region. This new inpainting feature mirrored the capabilities of Instruct pix2pix, offering users enhanced control over their editing processes.
September
🚀 AI Development
September was an exciting month, as OpenAI announced that ChatGPT can now “see, hear, and speak”. These enhanced capabilities of ChatGPT enabled users to to engage with the chatbot through voice commands or images, breaking free from the constraints of text-only interactions.
✏️ AI-Powered Content Creation
Content creators also got exciting news in September, as YouTube announced a new AI-enabled video editing app called YouTube Create, currently released as a beta version for select markets. The company also shared its current work on ‘Dream Screen’ - a new feature that would allow creators to seamlessly integrate AI-generated video or image backgrounds into their content.
🛠️ AI in Product Development
As curious as this may sound, September was also the month when Coca Cola introduced a new mystery flavor, created by AI. Coca-Cola Y3000 might indeed taste like the future, serving as a testament to AI’s creative potential in product development and marketing.
October
🚀 AI Development
In October, ChatGPT Plus and Enterprise users gained access to the company’s advanced image generation model, Dall-E 3. OpenAI touted DALL-E 3’s capability to render hands, text, and faces, areas where AI-powered photo generators often encounter challenges.
📷 AI Image & 📹 Video Generation
October also witnessed a notable expansion of Adobe’s suite of generative AI-powered tools with the introduction of Text-to-Vector in Adobe Illustrator.
This addition provided users with the ability to seamlessly convert text into vector graphics, enhancing the creative possibilities within the platform.
📜 AI Regulation
In a groundbreaking move at the White House, President Joe Biden took a decisive step toward securing the future by signing the US’s first ever Executive Order directed towards AI. The game-changing directive requires AI developers to open the curtain on their safety tests, sharing crucial results with the US government before going public.
By enforcing this measure, the administration aims to bolster accountability and ensure the responsible development and deployment of AI technologies for the benefit of society.
November
🚀 AI Development
November was also a month full of novelties and exciting news:
- OpenAI introduced GPTs — custom versions of ChatGPT that combine instructions, extra knowledge, and any combination of skills.
- Elon Musk rolled out chatbot Grok, his own LLM trained on available data from Twitter (now X).
- Microsoft made Microsoft 365 Copilot generally available for enterprise customers worldwide.
📰 AI News
November also brought unsettling news for OpenAI’s CEO, Sam Altman, who was abruptly fired from the company, only to be reinstated a few days later. Rumors suggest that a lack of transparency may have influenced OpenAI’s board to make this decision. However, the details surrounding this power struggle remain shrouded in secrecy, leaving observers to speculate about what truly transpired.
📷 AI Image & 📹 Video Generation
The AI artworld also experienced several exciting developments in November:
- StabilityAI launched Stable Video Diffusion, an innovative model for generating videos, built on Stable Diffusion.
- Runway Gen-2 introduced Motion Brush, an editing tool enabling users to add movement to specific parts of an image.
- Pika 1.0 was unveiled, heralding a game-changing platform in AI video generation that empowers users to create captivating videos using text prompts.
💡 AI & New Technologies
On November 29th, Google published a press release unveiling that Google DeepMind’s team used AI tool GNoMe to discover 2.2 million new crystals, including 380,000 stable materials that could power future technologies. This was groundbreaking news, showcasing AI’s potential to lead the way in materials discovery, experimentation, and synthesis, paving the path for innovative advancements in various industries.
🚑 AI in Healthcare
Continuing the trajectory of research breakthroughs, November also brought news of a significant advancement in breast cancer detection. Kheiron Medical Technologies and Imperial College London unveiled the discovery of a new AI tool named Mia, which has shown the potential to enhance early detection rates of breast cancer by up to 13%. While ongoing research is essential to validate these findings, this development marks a crucial stride toward enhancing cancer screening methodologies.
December
🚀 AI Development
In December, Google first introduced Gemini, their largest and most capable AI model to date, boasting state-of-the-art performance across many leading benchmarks.
Designed to be multimodal, Gemini demonstrates the ability to seamlessly comprehend, manipulate, and combine various forms of information, including text, code, audio, image, and video.
📜 AI Regulation
In December, significant strides were made in AI regulation and Responsible AI development, capturing global attention with several notable headlines:
- The European Union reaches a landmark agreement on the AI Act, marking a significant step forward in regulating AI technologies.
- IBM and Meta joined forces to launch the AI Alliance for Responsible AI development
2023: The Year When AI Safety and Transparency Emerged As Priorities
The past year also marked a pivotal turning point when AI safety and transparency emerged as urgent priorities. While 2023 witnessed an unparalleled surge in AI enthusiasm, it also served as a wake-up call for the global community regarding the critical importance of trust and transparency in AI systems.
AI isn’t new, but neither is AI risk. The AI Incident Database which chronicles incidents dating back to 1983, now contains 629 incidents and 3309 incident reports.
Without proper guardrails, we may well be speeding towards a new AI winter.
However, it’s essential to acknowledge that there is no universal solution to tackle these challenges. Navigating through this complex terrain requires a degree of experimentation, as regulators and policymakers worldwide endeavor to chart the best course ahead.
AI regulation and policy have progressed in 2023 and we believe that to be a good thing. Striking the balance between safety and innovation is paramount to harnessing the full potential of AI for societal benefit.
Make Responsible AI a Part of Your Company’s DNA
At Lumenova AI, we firmly believe in the tremendous potential of AI while also recognizing the significance of responsible AI practices.
From day one, our goal was to assist organizations like yours in automating, simplifying, and streamlining their entire AI risk management and compliance process, making AI Governance a seamless and efficient journey.
Whether your organization operates in healthcare, government, financial services, or any other sector, our AI Governance platform offers comprehensive support to drive successful business transformation.
We’re enthusiastic about a future where AI ethics seamlessly integrate with AI innovation and if you share our views, we’d love to show you how Lumenova AI works. Get in touch with us for a demo.