Interested in a hands-on learning experience for developing LLM applications?
Join our LLM Bootcamp today and Get 5% Off for a Limited Time!

Game-changing Advancements in AI: Rewinding 2023

December 31, 2023

2023 marked a pivotal year for advancements in AI, revolutionizing the field. We saw a booming architecture around this field, promising us a future filled with greater productivity and automation.

OpenAI took the lead with its powerful LLM-powered tool called ChatGPT, which created a buzz globally. What followed was unexpected. People started to rely on this tool as much as they rely on the internet.

This attracted the interest of big tech companies. We saw companies like Microsoft, Apple, Google, and more fueling this AI race.

Moreover, there was also a rise in the number of startups creating generative AI tools and building on to the technology around it. In 2023, investment in generative AI startups reached about $27 billion.

Long story short, generative AI proved to us that it is going to prevail. Let’s examine some pivotal events of 2023 that were crucial.

1. Microsoft and OpenAI’s announcement of the third phase of their partnership

Microsoft concluded the third phase of its strategic partnership with OpenAI, involving a substantial multibillion-dollar investment to advance AI breakthroughs globally.

Following earlier collaborations in 2019 and 2021, this agreement focused on boosting AI supercomputing capabilities and research. Microsoft increased investments in supercomputing systems and expanded Azure’s AI infrastructure.

The partnership aimed to democratize AI, providing broad access to advanced infrastructure and models. Microsoft deployed OpenAI’s models in consumer and enterprise products, unveiling innovative AI-driven experiences.

The collaboration, driven by a shared commitment to trustworthy AI, aimed to parallel historic technological transformations

Read more here

2. Google forged a partnership with Anthropic to deliver responsible AI

Google Cloud announced a partnership with the AI startup, Anthropic. Google Cloud was cemented as Anthropic’s preferred provider for computational resources, and they committed to building large-scale TPU and GPU clusters for Anthropic.

These resources were leveraged to train and deploy Anthropic’s AI systems, including a language model assistant named Claude.

Read more here

 

3. Google released its AI tool “Bard”

Google made a significant stride in advancing its AI strategy by publicly disclosing Bard, an experimental conversational AI service. Utilizing a vast trove of internet information, Bard was engineered to simplify complex topics and generate timely responses, a development potentially representing a breakthrough in human-like AI communication.

Read more about ChatGPT vs Bard

This announcement followed Google’s intent to make their language models, LaMDA and PaLM, publicly accessible, thereby establishing its commitment to transparency and openness in the AI sphere.

These advancements were part of Google’s response to the AI competition triggered by OpenAI’s launch of ChatGPT, exemplifying a vibrant dynamic in the global AI landscape that is poised to revolutionize our digital interactions moving forward.

 

 

Learn to build custom large language model applications today!                                                

 

4. Microsoft launched a revised Bing search powered by AI

Microsoft set a milestone in the evolution of AI-driven search technology by unveiling a revamped version of Bing, bolstered by AI capabilities. This integrated ‘next generation’ OpenAI model, regarded as more advanced than ChatGPT, is paired with Microsoft’s proprietary Prometheus model to deliver safer, more pertinent results.

Microsoft’s bold move aimed to scale the preview to millions rapidly and seemed designed to capture a slice of Google’s formidable search user base, even as it sparked fresh conversations about potential risks in AI applications.

 

5. Github Copilot for business became publicly available

GitHub made headlines by offering its AI tool, GitHub Copilot for Business, for public use, showcasing enhanced security features.

With the backing of an OpenAI model, the tool was designed to improve code suggestions and employ AI-based security measures to counter insecure code recommendations. However, alongside these benefits, GitHub urged developers to meticulously review and test the tool’s suggestions to ensure accuracy and reliability.

The move to make GitHub Copilot publicly accessible marked a considerable advancement in the realm of AI-powered programming tools, setting a new standard for offering assistive solutions for coders, even as it underscored the importance of vigilance and accuracy when utilizing AI technology.

Further illustrating the realignment of resources towards AI capabilities, GitHub announced a planned workforce reduction of up to 10% by the end of fiscal year 2023.

6. Google introduced two generative AI capabilities to its cloud services, Vertex AI and Generative AI App Builder

Google made a substantial expansion of its cloud services by introducing two innovative generative AI capabilities, Vertex AI and Generative AI App Builder. The AI heavyweight equipped its developers with powerful tools to harness AI templates for search, customer support, product recommendation, and media creation, thus enriching the functionality of its cloud services.

These enhancements, initially released to the Google Cloud Innovator community for testing, were part of Google’s continued commitment to make AI advancements accessible while addressing obstacles like data privacy issues, security concerns, and the substantial costs of large language model building.

 

8. AWS launched Bedrock

Amazon Web Services unveiled its groundbreaking service, Bedrock. Bedrock offers access to foundational training models from AI21 Labs, Anthropic, Stability AI, and Amazon via an API. Despite the early lead of OpenAI in the field, the future of generative AI in enterprise adoption remained uncertain, compelling AWS to take decisive action in an increasingly competitive market.

As per Gartner’s prediction, generative AI is set to account for 10% of all data generated by 2025, up from less than 1% in 2023. In response to this trend, AWS’s innovative Bedrock service represented a proactive strategy to leverage the potential of generative AI, ensuring that AWS continues to be at the cutting edge of cloud services for an evolving digital landscape

 

9. OpenAI released Dall. E 2

OpenAI launched an improved version of its cutting-edge AI system, DALL·E 2. This remarkable analytic tool uses AI to generate realistic images and artistry from textual descriptions, stepping beyond its predecessor by generating images with 4x the resolution.

It also expand images beyond the original canvas. Safeguards were put in place to limit the generation of violent, hateful, or adult images, demonstrating its evolution in responsible AI deployment. Overall, DALL·E 2 represented an upgraded, more refined, and more responsible version of its predecessor.

 

10. Google increased Bard’s ability to function as a programming assistant

Bard became capable of aiding in critical development tasks, including code generation, debugging, and explaining code snippets across more than 20 programming languages. Google’s counsel to users to verify Bard’s responses and examine the generated code meticulously spoke to the growing need for perfect programming synergies between AI and human oversight.

Despite potential challenges, Bard’s unique capabilities paved the way for new methods of writing code, creating test scenarios, and updating APIs, strongly underpinning the future of software development.

 

Learn how Generative AI is reshaping the world and future as we know it. Watch our podcast Future of Data and AI now.

11. The White House announced a public evaluation of AI systems

The White House announced a public evaluation of AI systems at the DEFCON 31 gathering in Las Vegas.

This call resonated with tech leaders from powerhouses, such as Alphabet, Microsoft, Anthropic, and OpenAI, who solidified their commitment to participate in the evaluation, signaling a crucial step towards demystifying the intricate world of AI.

In conjunction, the Biden administration announced its support by declaring the establishment of seven new National AI Research Institutes, backed by an investment of $140 million, promising further growth and transparency around AI.

This declaration, coupled with the commitment from leading tech companies, held critical significance by creating an open dialogue around AI’s ethical use and promising regulatory actions toward its safer adoption

 

12. ChatGPT Plus can browse the internet in beta mode

ChatGPT Plus announced the beta launch of its groundbreaking new features, allowing the system to navigate the internet.

This feature empowered ChatGPT Plus to provide current and updated answers about recent topics and events, symbolizing a significant advance in generative AI capabilities.

Wrapped in user intrigue, these features were introduced through a new beta panel in user settings, granting ChatGPT Plus users the privilege of early access to experimental features that could change during the developmental stage.

 

13. OpenAI rolled out code interpreter

OpenAI made an exciting announcement about the launch of the ChatGPT Code Interpreter. This new plugin was a gift to all the ChatGPT Plus customers that would roll out to them over the next week. With this plugin, ChatGPT expanded its horizon by giving a new way of executing Python code within the chatbot interface.

The code interpreter feature wasn’t just about running the code. It brought numerous promising capabilities, like carrying out data analysis, managing file transfers, and even the chance to modify and improve code. However, the only hitch was that one couldn’t use multiple plugins at the same time.

 

14. Anthropic released Claude-2

Claude 2, Anthropic AI’s latest AI chatbot, is a natural-language-processing conversational assistant designed for various tasks, such as writing, coding, and problem-solving.

Notable for surpassing its predecessor in educational assessments, Claude 2 excels in performance metrics, displaying impressive results in Python coding tests, legal exams, and grade-school math problems.

Its unique feature is the ability to process lengthy texts, handling up to 100,000 tokens per prompt, setting it apart from competitors.

 

14. Meta released open source model, Llama 2

Llama 2 represented a pivotal step in democratizing access to large language models. It built upon the groundwork laid by its predecessor, LLaMa 1, by removing noncommercial licensing restrictions and offering models free of charge for both research and commercial applications.

This move aligned with a broader trend in the AI community, where proprietary and closed-source models with massive parameter counts, such as OpenAI’s GPT and Anthropic’s Claude, had dominated.

Noteworthy was Llama 2’s commitment to transparency, providing open access to its code and model weights. In contrast to the prevailing trend of ever-increasing model sizes, Llama 2 emphasized advancing performance with smaller model variants, featuring seven billion, 13 billion, and 70 billion parameters.

 

15. Meta introduced Code Llama

Code Llama, a cutting-edge large language model tailored for coding tasks, was unveiled today. Released as a specialized version of Llama 2, it aimed to expedite workflows, enhance coding efficiency, and assist learners.

Supporting popular programming languages, including Python and Java, the release featured three model sizes—7B, 13B, and 34B parameters. Additionally, fine-tuned variations like Code Llama – Python and Code Llama – Instruct provided language-specific utilities.

With a commitment to openness, Code Llama was made available for research and commercial use, contributing to innovation and safety in the AI community. This release is expected to benefit software engineers across various sectors by providing a powerful tool for code generation, completion, and debugging.

 

16, OpenAI launched ChatGPT enterprise

OpenAI launched an enterprise-grade version of ChatGPT, its state-of-the-art conversational AI model. This version was tailored to offer greater data control to professional users and businesses, marking a considerable stride towards incorporating AI into mainstream enterprise usage.

Recognizing possible data privacy concerns, one prominent feature provided by OpenAI was the option to disable the chat history, thus giving users more control over their data. Striving for transparency, they also provided an option for users to export their ChatGPT data.

The company further announced that it would not utilize end-user data for model training by default, displaying its commitment to data security. If chat history was disabled, the data from new conversations was stored for 30 days for abuse review before when it was permanently deleted

 

17. Amazon invested $4 billion in Anthropic

Amazon announced a staggering $4 billion investment in AI start-up Anthropic. This investment represented a significant endorsement of Anthropic’s promising AI technology, including Claude 2, its second-generation AI chatbot.

The financial commitment was a clear indication of Amazon’s belief in the potential of Anthropic’s AI solutions and an affirmation of the e-commerce giant’s ambitions in the AI domain.

To strengthen its position in the AI-driven conversational systems market, Amazon paralleled its investment by unveiling its own AI chatbot, Amazon Q.

This significant financial commitment by Amazon not only emphasized the value and potential of advanced AI technologies but also played a key role in shaping the competitive landscape of the AI industry.

 

18. President Joe Biden signed an executive order for Safe AI

President Joe Biden signed an executive order focused on ensuring the development and deployment of Safe and Trustworthy AI.

President Biden’s decisive intervention underscored the vital importance of AI systems adhering to principled guidelines involving user safety, privacy, and security.

Furthermore, the move towards AI regulation, as evinced by this executive order, indicates the growing awareness and acknowledgment at the highest levels of government about the profound societal implications of AI technology.

 

19. OpenAI released its multimodal model, GPT-4 Vision and Turbo

OpenAI unveiled GPT-4 Turbo, an upgraded version of its GPT-4 large language model, boasting an expanded context window, increased knowledge cutoff to April 2023, and enhanced pricing for developers using the OpenAI API. Notably, “GPT-4 Turbo with Vision” introduced optical character recognition, enabling text extraction from images.

The model was set to go multi-modal, supporting image prompts and text-to-speech capabilities. Function calling updates streamlined interactions for developers. Access was available to all paying developers via the OpenAI API, with a production-ready version expected in the coming weeks.

 

20. Sam Altman was fired from OpenAI and then rehired in 5 days

OpenAI experienced a tumultuous series of events as CEO Sam Altman was abruptly fired by the board of directors, citing a breakdown in communication. The decision triggered a wave of resignations, including OpenAI president Greg Brockman.

However, within days, Altman was reinstated, and the board was reorganized. The circumstances surrounding Altman’s dismissal remain mysterious, with the board stating he had not been “consistently candid.”

The chaotic events underscore the importance of strong corporate governance in the evolving landscape of AI development and regulation, raising questions about OpenAI’s stability and future scrutiny.

 

21. Google released its multimodal model called Gemini

Gemini, unveiled by Google DeepMind, made waves as a groundbreaking AI model with multimodal capabilities, seamlessly operating across text, code, audio, image, and video. The model, available in three optimized sizes, notably demonstrates state-of-the-art performance, surpassing human experts in massive multitask language understanding.

Gemini excels in advanced coding, showcasing its proficiency in understanding, explaining, and generating high-quality code in popular programming languages.

With sophisticated reasoning abilities, the model extracts insights from complex written and visual information, promising breakthroughs in diverse fields. Its past accomplishments position Gemini as a powerful tool for nuanced information comprehension and complex reasoning tasks.

 

22. The European Union put forth its first AI Act

The European Union achieved a historic milestone with the adoption of the AI Act, the world’s inaugural comprehensive AI law, influencing global AI governance. The act, now a key moment in regulating artificial intelligence, classified AI systems based on risk and prohibited certain uses, ensuring a delicate balance between innovation and safety. It emphasized human oversight, transparency, and accountability, particularly for high-risk AI systems.

The legislation mandated stringent evaluation processes and transparency requirements for companies, promoting responsible AI development. With a focus on aligning AI with human rights and ethical standards, the EU AI Act aimed to safeguard citizens, foster innovation and set a global standard for AI governance.

 

23. Amazon released its model “Q”

Amazon Web Services, Inc. unveiled Amazon Q, a groundbreaking generative artificial intelligence assistant tailored for the workplace. This AI-powered assistant, designed with a focus on security and privacy, enables employees to swiftly obtain answers, solve problems, generate content, and take action by leveraging data and expertise within their company.

 

Read more about Q* in this blog

 

Among the prominent customers and partners eager to utilize Amazon Q are Accenture, Amazon, BMW Group, Gilead, Mission Cloud, Orbit Irrigation, and Wunderkind. Amazon Q, equipped to offer personalized interactions and adhere to stringent enterprise requirements, marks a significant addition to the generative AI stack, enhancing productivity for organizations across various sectors.

 

The Future of Advancements in AI:

Throughout 2023, generative AI made striking progress globally, with several key players, including Amazon, Google, and Microsoft, releasing new and advanced AI models. These developments catalyzed substantial advancements in AI applications and solutions.

Amazon’s release of ‘Bedrock’ aimed at scaling AI-based applications. Similarly, Google launched Bard, a conversational AI service that simplifies complex topics, while Microsoft pushed its AI capabilities by integrating OpenAI models and improving Bing’s search capabilities.

Notably, intense focus was also given to AI and model regulation, showing the tech world’s rising awareness of AI’s ethical implications and the need for responsible innovation.

Overall, 2023 turned out to be a pivotal year that revitalized the race in AI, dynamically reshaping the AI ecosystem

 

Originally published on LinkedIn by Data Science Dojo

Data Science Dojo | data science for everyone

Discover more from Data Science Dojo

Subscribe to get the latest updates on AI, Data Science, LLMs, and Machine Learning.