The big tech companies are currently driving an incredible amount of innovation and development, making it difficult to keep up with the constant stream of new models and technologies. These companies frequently host major events to showcase their latest advancements.

Apple recently concluded its highly anticipated WWDC 23 event, while Microsoft wrapped up its Build 2023 event. Google also held its Google IO 2023 in March. These events serve as platforms for top AI executives to release a range of new products.

One noticeable trend is the rise of generative AI, which has captivated the attention of the community. People who were previously unfamiliar with AI and machine learning now have an increasing interest in these technologies, thanks to the wave of generative AI.

The big tech companies have all heavily invested in generative AI and AI/ML, shifting their focus from other deep learning methods. Noteworthy developments include Google AI, Microsoft Copilot, Apple Machine Learning, and OpenAI’s pursuit of AGI (Artificial General Intelligence).

Apple’s Machine Learning

Apple has made significant investments in machine learning research, assembling a talented team of researchers and engineers. They have applied machine learning to various projects, including Siri, Photos, Health, and CarPlay, enhancing user experiences. Apple’s long-term commitment to machine learning is evident in its Machine Learning Research Residency Program, which offers training to early-career researchers.

The company’s love for Machine Learning was also very evident at its recent WWDC 2023 event, as it steered clear from parroting the term ‘AI’ unlike all its competitors including Google.

In 2023, Apple introduced new machine learning-based features like Live Text, Visual Look Up, and Safety Check in iOS 16. These initiatives demonstrate Apple’s dedication to leveraging machine learning to transform user interactions and improve their products and services. Expect Apple to continue investing in machine learning research and developing new machine learning-driven offerings in the future.

Apple CEO Tim Cook has also positioned the company as uninterested in collecting user data, which he believes sets Apple apart from companies like Google and Facebook. However, this aversion to cloud computing poses a challenge as Apple seeks to develop new machine learning and AI-powered features. Building and running machine learning services require computing power and data, both of which are more readily available in the cloud. While Apple’s mobile devices have impressive capabilities, they may struggle to compete with servers, especially those equipped with Google’s custom machine learning chips.

GoogleAI

Google has been a significant player in AI research and development, with initiatives like Google Brain and programs such as the Google AI Residency Program. The company has made groundbreaking advancements in AI algorithms and systems, leading to the creation of AI-powered products and services like Google Search, Google Translate, and Google Photos. Google actively conducts and publishes AI research findings and invests in the potential of AI to address global challenges.

However, Google is now facing competition from OpenAI and Microsoft, particularly in the field of generative AI. At Google I/O, the focus was on Bard, a chatbot aimed at competing with OpenAI’s ChatGPT. Some experts feel that Google’s recent approach has been reactive and divergent from its innovation-focused past. The company has shifted its AI operations to prioritize quick product launches, which has led to concerns about neglecting its AI history and potentially falling behind in the market.

Google’s parent company, Alphabet, has been investing in AI for years and acquired DeepMind in 2014. Recently, Alphabet merged its Google Research team with DeepMind to consolidate AI efforts. However, some experts believe that this consolidation should have been done earlier, as Google experienced a “Kodak moment” by not capitalising on its leading AI product, falling behind Microsoft in 2022.

To strengthen its AI focus, Google has made investments in companies like Anthropic, showing its commitment to advancing AI technology. While Google’s prior investments and strong AI technology remain relevant, the company is working to catch up with competitors and bring AI into its products more quickly, similar to the strides made by Microsoft.

Microsoft and Copilot

Microsoft has invested heavily in artificial intelligence in recent years, and its Copilot project is one of the most ambitious examples of this investment. It’s a powerful language model that generates text, translates languages, and assists with various creative tasks. Copilot aims to transform how people work and create by enhancing productivity, fostering creativity, and promoting inclusivity. Microsoft plans to offer Copilot as a free service for Microsoft 365 subscribers and as a standalone product. The tool has the potential to revolutionise AI’s impact on the world, with benefits including increased productivity, improved quality, and expanded creativity. Microsoft has also expanded Copilot’s application in CRM and ERP with Dynamics 365 Copilot, and GitHub has launched Github Copilot for Business, an AI coding assistant for public use.

OpenAGI

OpenAI’s CEO, Sam Altman, and other founders have discussed artificial general intelligence (AGI) on various platforms, expressing both optimism about its potential benefits and concern for its risks. Altman stated in an interview with Lex Fridman that he believes AGI is “probably 10 to 20 years away” and could have a “positive impact on humanity,” emphasising the need to ensure its responsible use.

During his ongoing India visit as well, Altman is discussing AGI. He sees AGI as 10 to 20 years away with the potential to solve global problems. Altman acknowledges AGI risks, including misuse and job displacement. He believes India’s talent and population make it a potential AGI leader. Altman emphasises the importance of considering AGI’s risks and benefits now. He works on safety guidelines and builds an expert community for responsible AGI use. Altman’s visit reflects the growing interest in AGI worldwide. It’s crucial to contemplate AGI’s potential benefits and risks as it becomes more realistic. OpenAI, under Altman’s leadership, focuses on safe and ethical AGI development.

In a blog post, Altman and the other founders outlined their vision for AGI, stating that it could “solve some of the world’s most pressing problems” such as climate change, poverty, and disease, while also fostering human creativity and ingenuity.

However, they acknowledged the potential risks of AGI, including malicious use for creating autonomous weapons or causing mass unemployment by replacing human jobs.

Amazon and Cloud Services

Amazon has invested heavily in AI research and their cloud services serve as a prominent platform for AI development and deployment. Their AI research team focuses on enhancing the performance of cloud services through the development of new AI technologies.

Research areas include machine learning (ML), with a focus on algorithms and models for training and deploying ML models. This research improves the performance of Amazon’s cloud services like SageMaker, Forecast, and Personalise. Additionally, Amazon’s AI research team is dedicated to developing tools and resources for AI developers, which are accessible through their AI research website.

The cloud-based platform SageMaker enables the building, training, and deployment of ML models for various applications such as fraud detection, customer churn prediction, and product recommendations. Amazon’s AI research efforts are advancing the capabilities and versatility of their cloud, opening up new possibilities for businesses and developers to leverage AI in their operations and products.

The recently released Falcon 40B, a large language model, is developed on Amazon Web Services (AWS). The Falcon 40B is a versatile and robust tool for translation, question answering, summarisation, and image identification and is accessible on AWS through Amazon SageMaker JumpStart.

Meta and Self-Supervised Learning

Meta began its SSL journey in 2017, exploring its potential to improve machine learning performance. They developed SSL methods like SimCLR, SwAV, and DINO, achieving state-of-the-art results in tasks like image classification and object detection. Meta invested in large-scale compute clusters, enabling the training of significantly larger SSL models. This progress has had a major impact on AI, with SSL widely used and considered a promising approach. Key milestones include the introduction of SimCLR in 2018, SwAV in 2019, and DINO in 2020. Meta built Megatron, a compute cluster for SSL training, in 2021. In 2022, they published the Data2vec paper, introducing an SSL algorithm across speech, vision, and text modalities. Meta’s continued investment in SSL research will lead to further advances.

As Meta’s VP & Chief AI Scientist Yann LeCunn has emphasised repeatedly, he doesn’t believe in RLHF and thinks it is, “I think RLHF is hopeless because the space of wrong answers is very large, and the space of tricky questions has a very long tail. Any system that does not experience the world and learn for itself is going to be at the mercy of the data it is given to learn from.”