Generative AI: The Biggest Stories of 2024

1. NVIDIA AI Architecture Sold Out

NVIDIA dominated the AI space in 2024, introducing the Blackwell chip in March. This revolutionary microarchitecture enabled AI training, research, and computing for top companies like Amazon Web Services, Google, Meta, Microsoft, OpenAI, and xAI. By October, Blackwell chips had sold out for the following year, as companies scrambled to get their hands on these powerful processors.

In March, NVIDIA’s Hopper chips helped the company reach $2 trillion in market capitalization, making it one of the three most valuable companies worldwide, alongside Microsoft and Apple. AMD and Intel also provide AI accelerator chips, but their businesses haven’t experienced the same explosive growth as NVIDIA’s.

The success of these powerful processors is just one of many ripple effects driven by the growing demand for larger, higher-density data centers within the broader generative AI industry.

2. OpenAI Showed Its Secretive o1

Many major AI companies moved forward with enterprise offerings, more powerful models, and experimented with new hardware in 2024. For OpenAI, whispers circulated about a “Strawberry” model that would take the next leap toward humanlike intelligence. However, it turned out to be OpenAI o1, a “reasoning” model designed to tackle more complex problems than its predecessors in the GPT-4 family.

3. The AI PC Became Mainstream

Tech enthusiasts may reflect on 2024 as the year AI became a standard feature in nearly every new PC. From Apple Intelligence to Microsoft Copilot, built-in AI was everywhere. In September 2024, Gartner predicted AI PCs would make up 43% of all PC shipments by 2025.

However, a November article in Reuters found that PC demand remained low overall – although NVIDIA’s AI processors being on backorder may have limited the availability of AI PCs.

4. Did AI Bring a New Way of Thinking About UI?

Since the debut of the Apple Store in 2008, apps have been the primary way consumers interact with smart devices. Voice assistants like Siri have added another convenience layer, enabling users to control certain applications through voice commands. The AI industry wants the next stage to be seamless control of all aspects of a PC or phone through AI, as demonstrated by Microsoft Copilot and Claude 3.5 Sonnet’s Computer Use feature.

Computer Use with Claude allows AI to translate natural language instructions into actionable commands, such as moving cursors, typing, and interacting with a computer like a human would. However, this functionality can be resource-intensive. For example, performing a simple task like opening a URL and extracting information from a website can cost as much as $0.31 in tokens.

5. Microsoft Recall Was Delayed Repeatedly

Microsoft Recall sparked controversy from the outset due to its unlimited access between Copilot and the rest of the PC, raising security concerns. Once planned for a public preview in June, early access to Recall was delayed until October and then December. Recall was intended to be a cornerstone of Microsoft AI PCs, but Microsoft’s delays in pursuit of “a secure and trusted experience” show that Recall pushed the boundaries of what consumers are willing to share with AI.

6. ChatGPT Adds ‘Search’ Option

In October, OpenAI expanded ChatGPT with a Google-like search engine functionality. ChatGPT search provides generative answers with links to external sites. Adding up-to-date information like weather reports could give OpenAI a competitive edge over Google search. OpenAI made deals with several media outlets to license their content to appear in ChatGPT searches.

Initially, only ChatGPT Plus and Team users had access to ChatGPT search. However, in December, the tech giant made it available to all users.

7. Apple Intelligence Launched

Apple kept quiet during much of the AI race, waiting until 2024 to unveil its plans. Apple Intelligence was announced in June for newer devices. In cooperation with OpenAI, Apple added many standard AI capabilities, such as summarization and rewriting. It also brings limited image creation, mostly in cartoony styles to avoid the possibility of users making deepfakes.

Apple Intelligence uses the tech giant’s M-series chips or the A17 Pro chip or later. As of iOS 18.2, Apple devices can connect to ChatGPT to handle more complex questions posed to Apple Intelligence or Siri.

8. Google Explored Gemini Use Cases for Enterprise

Google Gemini wasn’t a new development for 2024. However, its launch in December 2023 and the release of a smaller model, Gemma, in February means most of Gemini’s public life happened this year. Google replaced its Bard brand with the more powerful model, bringing Gemini to search, mobile apps, Chromebooks, and the Vertex AI cloud platform. Following the “agentic” AI trend, Google released custom “Gems” in August.

9. AI Regulation Continued to Develop

In 2024, governments worked to regulate AI use. In Europe, the European Union’s AI Act went into effect in August, outlining banned uses while trying to provide guidelines to foster innovation.

The U.K. created an AI assurance market to pursue business in the generative AI sector, with more legislation expected next year. The U.K. joined several international initiatives with the U.S. and others to standardize AI safety.

In the U.S., the Federal Trade Commission (FTC) has been actively pursuing AI-related enforcement actions, focusing on issues like deceptive advertising and data privacy.

10. Video Generation Technology Matured

Generative AI video remains imperfect, often producing uncanny results with inconsistent scenes and oddly proportioned or distorted limbs. However, that hasn’t stopped companies from releasing AI video generators.

OpenAI’s Sora, first demonstrated in February, was released to ChatGPT Plus and Pro users in December. Google’s Veo is available to select Google Cloud customers. Even Canva offers AI video generation.

Conclusion

2024 was a transformative year for generative AI, with numerous breakthroughs and innovations across various industries. From AI architecture to video generation technology, the field continues to evolve at an incredible pace. As AI becomes more ubiquitous, it’s essential to consider the ethical implications and regulatory frameworks that will shape its future development.

FAQs

Q: What is the Blackwell chip?
A: The Blackwell chip is a revolutionary microarchitecture introduced by NVIDIA in March 2024, enabling AI training, research, and computing for top companies.

Q: What is OpenAI o1?
A: OpenAI o1 is a “reasoning” model designed to tackle more complex problems than its predecessors in the GPT-4 family.

Q: What is Microsoft Recall?
A: Microsoft Recall is a feature that allows AI to translate natural language instructions into actionable commands, such as moving cursors, typing, and interacting with a computer like a human would.

Q: What is ChatGPT search?
A: ChatGPT search is a Google-like search engine functionality added to ChatGPT in October 2024, providing generative answers with links to external sites.

Q: What is Apple Intelligence?
A: Apple Intelligence is a feature announced in June 2024 for newer devices, adding many standard AI capabilities, such as summarization and rewriting.

Q: What is Google Gemini?
A: Google Gemini is a model that was launched in December 2023 and released a smaller model, Gemma, in February 2024, bringing Gemini to search, mobile apps, Chromebooks, and the Vertex AI cloud platform.

Q: What is AI regulation?
A: AI regulation refers to the efforts of governments to regulate AI use, focusing on issues like deceptive advertising, data privacy, and ethics.