My 2024 AI & GenAI Prediction Scorecard: What I Got Right (and Wrong) This Year

Just over a year ago in 2023, I shared a set of bold predictions for where AI—and especially Generative AI—would be in 2024.

2024 AI and Gen AI Predictions

Now that the time has come, it’s time to see how my 2024 AI and GenAI predictions held up. Spoiler alert: some were spot-on, others missed the mark, and—like any year in tech—there were plenty of surprises I never saw coming. Let’s dive in and explore what actually happened in the world of AI.

Percentage of Correct 2024 Predictions:

Overall, looking at the state of AI as we enter (or near) 2024, around 65% of my 2024 AI and GenAI predictions were right or on track. While some predictions (like the continued rise of open-source models and growth of on-device AI) are spot-on, others (e.g., Apple fully replacing Siri or launching a direct Google competitor) haven’t quite materialized—or at least not yet.

1. Apple will beta release a Google Competitor

• What I Predicted: Apple would roll out a search engine or technology rivaling Google’s dominance.
• What Happened: While Apple continues to invest heavily in AI under the hood, there hasn’t been a direct, public “Google competitor.” Rumors swirl about Apple GPT and more AI-driven Spotlight and Siri enhancements, but a full-fledged search engine isn’t here yet.

2. Apple will replace Siri

• What I Predicted: Siri, as we know it, would be replaced by a more advanced AI assistant.
• What Happened: Siri has seen incremental improvements, but it hasn’t been completely replaced. Apple’s focus on new AI features (like on-device machine learning) suggests bigger changes might be coming—just not yet in 2024.

3. Targeted and Smaller Language Models

• What I Predicted: We’d see LLMs that are more specialized and more compact.
• What Happened: This has largely come true. We’ve witnessed a shift toward smaller, domain-specific models that organizations can deploy more efficiently. This is a big win for privacy and cost-effectiveness.

4. On-Device: 2024 and beyond will be all about Smaller Models “on-device”

• What I Predicted: AI would increasingly move from large cloud-hosted models to smaller, on-device deployments.
• What Happened: Indeed, there’s a growing trend toward running LLMs locally (or at least partly on-device), especially on smartphones and specialized hardware. Advances in chips and edge computing are making it feasible.

5. We will see more Multimodal Large Language Models

• What I Predicted: LLMs that process multiple types of data—text, images, audio, video—would multiply.
• What Happened: GPT-4, PaLM-E, Llama 2, and other models have made strides in multimodality. This is an area that’s expanding quickly, so the prediction about more multimodal approaches was definitely correct.

6. Large Language Models will be moving into the Domain of Perception

• What I Predicted: LLMs would expand into perceiving and interpreting visual and audio signals.
• What Happened: We’re seeing LLMs integrated with APIs for image recognition, text-to-speech, and other sensory inputs. Though not fully “human-like” in perception, the forward steps are undeniable.

7. LLMs will move into World Modeling

• What I Predicted: Models would handle more contextual understanding of the world, bridging real-time data and logical inference.
• What Happened: We do see some advanced AI systems combining language with real-world data (for instance, ChatGPT plugins). True, comprehensive “world modeling” is still emerging—but the seeds are definitely there.

8. AI-generated video will improve in quality exponentially

• What I Predicted: AI video generation would get dramatically better.
• What Happened: Tools like Runway, Pika Labs, and others have shown huge improvements. While “exponential” might be a bit strong, the advances in video quality and speed are evident.

9. New Assistive AI and GenAI products for disabilities will be developed

• What I Predicted: AI would power next-gen accessibility tools for people with disabilities.
• What Happened: We’ve seen new assistive products that leverage speech-to-text, text-to-speech, and multimodal capabilities. There’s still enormous potential here, and 2024 has indeed offered more solutions than ever before.

10. Multimodal AI: Will increase productivity in many industries

• What I Predicted: Combining different data types in AI models would boost productivity.
• What Happened: From retail to healthcare, companies use multimodal AI to streamline tasks, identify patterns, and cut costs. This prediction holds true across many sectors.

11. Generative AI video: Quality will reach Hollywood standards

• What I Predicted: AI-generated video could rival major studio productions.
• What Happened: While the gap between AI-generated and Hollywood-level productions is shrinking, it’s still noticeable. Skilled creators are using advanced AI for pre-visualization, concept art, and short-form content, but “Hollywood standards” are still a tall order for purely AI-driven video.

12. Responsible AI: 2024 will be the year of responsible AI

• What I Predicted: A big focus on transparency, governance, and ethical use.
• What Happened: Regulatory bodies worldwide have turned up the heat on data use and AI ethics. With new guidelines, frameworks, and governmental oversight, responsible AI truly has taken center stage.

13. Open Source Models: Will continue to lead the way for broader adoption of generative AI

• What I Predicted: Open source frameworks would spur widespread adoption.
• What Happened: Projects like Llama 2, Falcon, and many others are fueling AI progress outside the walled gardens. This is spot-on—open source is thriving, and it’s allowing more developers and businesses to innovate.

14. AI Data – Companies need better data to stand out

• What I Predicted: High-quality datasets would become a key differentiator, and big players might even acquire media companies for data.
• What Happened: Curated, proprietary data is increasingly crucial. While we haven’t seen major media conglomerates snatched up solely for training data yet, deals and partnerships focusing on content libraries for AI training have picked up.

15. AI-enabled assistants: Many people will have an AI-enabled assistant by the end of 2024

• What I Predicted: Widespread adoption of personal AI.
• What Happened: Assistants like ChatGPT, Bing Chat, and others are definitely more integrated into our routines. We might not yet all have a single “AI assistant,” but the presence of AI in daily life is hard to miss.

16. AI Agents: Will take off in 2024

• What I Predicted: Agents tasked with specific goals would see massive growth.
• What Happened: Auto-GPT, AgentGPT, and numerous specialized solutions (marketing agents, coding agents, etc.) have made waves. Though still somewhat niche, the trajectory looks strong.

17. AI ‘Crews’: Multiple AI Agents strung together, with Tasks and Processes

• What I Predicted: A chain-of-agents approach—where multiple agents collaborate—would emerge.
• What Happened: We are seeing experiments in orchestrating multiple models (or “agents”) to tackle complex workflows. While it hasn’t gone entirely mainstream, the concept of multi-agent frameworks is on the rise.

18. AI Marketplaces & AI App Stores

• What I Predicted: AI would reinvent marketplaces like Fiverr or Upwork, and we’d see a surge in new AI “app stores.”
• What Happened: We’ve seen some early disruption with AI-powered freelancers and specialized AI services. OpenAI’s GPTs (or “chatbot templates”) and aggregator platforms for AI apps are popping up, but it’s still early days.

19. 2024 will be about creating partnerships and collaborations

• What I Predicted: Cloud providers, data storage, and analytics companies would form strategic alliances with AI startups and enterprises.
• What Happened: AWS, Azure, and Google Cloud continue to partner with AI ventures at lightning speed. We’ve witnessed new integrations, co-selling agreements, and specialized programs to accelerate enterprise AI.

20. 2024 will be the first full year we see how consumers respond to AI-creative works

• What I Predicted: Consumer acceptance and spending on AI-driven creative content would become a clear indicator of AI’s future.
• What Happened: From AI-generated music to short films, the consumer reaction is mixed but growing. Some audiences are fascinated, others more cautious. It’s still an emerging market, but adoption is inching forward.

21. Multimodal AI: Will increase productivity in many industries

• What I Got Right: Multimodal AI definitely expanded in 2024, combining text, image, audio, and even video capabilities. Companies in healthcare, marketing, and beyond used multimodal solutions to streamline processes and improve user experiences.
• What Didn’t Come True: Not every industry adopted multimodal AI at the same pace. While uptake is significant, “productivity boosts” vary widely depending on organizational readiness and data quality.

22. Generative AI video: Quality will reach Hollywood standards, used in interesting ways by skilled creators

• What I Got Right: Generative video tools improved notably, allowing creators to generate short clips and special effects that would have been unthinkable just a couple of years ago. Indie filmmakers and marketing teams quickly jumped on these tools.
• What Didn’t Come True: True “Hollywood standards” remain elusive for purely AI-generated video. High-budget studios still rely on massive VFX teams for blockbuster-level results, although AI is increasingly used for pre-visualization and concept work.

23. Responsible AI: 2024 will be the year of responsible AI

• What I Got Right: Regulatory bodies and industry coalitions worldwide are focusing on ethical AI use, transparency, and data governance. Many businesses have adopted frameworks and guidelines for deploying AI responsibly.
• What Didn’t Come True: While progress was made, some regions and industries have lagged behind in establishing robust oversight or consistent enforcement. “The year of responsible AI” might be a bit strong since the conversation is still evolving.

24. Open Source Models: Will continue to lead the way for broader adoption of generative AI

• What I Got Right: Projects like Llama 2 and other open-source models have exploded in popularity, enabling developers and startups to innovate without the high costs associated with proprietary platforms.
• What Didn’t Come True: Fully “leading the way” might be an overstatement. Proprietary models from major tech players still dominate in many enterprise settings, but open-source undeniably accelerated adoption and experimentation.

25. AI Data – AI companies need to secure better data to stand out

• What I Got Right: The importance of high-quality, proprietary datasets has become a clear differentiator, with many companies investing heavily in unique data acquisition and curation.
• What Didn’t Come True: While there’s talk about acquiring media companies for their content, we haven’t seen blockbuster buyouts of major media properties solely for training data. Partnerships and licensing deals are more common than outright acquisitions.

26. AI-enabled assistants: Many people will have an AI-enabled assistant by the end of 2024

• What I Got Right: Personal AI tools—like ChatGPT, Bing Chat, and various specialized assistants—are ubiquitous now, integrated into daily workflows and personal routines.
• What Didn’t Come True: Not everyone relies on a single “intelligent assistant” for all tasks. Instead, many people juggle multiple AI tools (e.g., for writing, scheduling, brainstorming) rather than one unified assistant.

27. AI Agents: Will take off in 2024

• What I Got Right: Specialized agents (e.g., Auto-GPT, AgentGPT) and sector-specific agents (like sales bots or coding assistants) are increasingly used in enterprise settings.
• What Didn’t Come True: While interest in agents has grown, many solutions remain experimental or limited to tech-savvy users, rather than becoming mainstream in every organization.

28. AI 'Crews': Multiple AI Agents (Multi-Agents) strung together with Tasks and Processes

• What I Got Right: The concept of chaining multiple AI models or “agents” together to tackle complex tasks is gaining traction, especially in large enterprises that can afford to experiment.
• What Didn’t Come True: This remains more of a cutting-edge approach than a widely adopted practice. Implementations tend to be proof-of-concepts or niche deployments rather than standard operating procedure.

29. AI Marketplaces: AI will reinvent marketplaces like Fiverr, Preply, Upwork, etc.

• What I Got Right: These platforms are starting to incorporate AI offerings—like freelance AI services or tutoring bots—giving rise to new ways of matching skills and tasks.
• What Didn’t Come True: A total “reinvention” hasn’t happened yet. Most established marketplaces still center on human freelancers, though some are experimenting with AI-assisted or AI-only listings.

30. AI App Stores: A shift from infrastructure to application layers, with thousands of new AI apps emerging

• What I Got Right: The number of AI applications, plugins, and specialized tools has soared. OpenAI GPTs, in particular, spawned a vibrant ecosystem of AI-driven mini-apps.
• What Didn’t Come True: We haven’t quite seen a “gold rush” on the scale of the mobile app explosion yet. Many AI app offerings remain bundled within larger platforms or cloud marketplaces rather than standing fully on their own.

31. Generative AI music models: Will be used in major productions including games

• What I Got Right: Indie and mid-tier game developers, as well as content creators, are experimenting with AI-generated soundtracks and effects.
• What Didn’t Come True: Major AAA game studios often still rely on human composers to deliver consistently high-quality, emotionally resonant music, though AI can be used for ideation or filler tracks.

32. Generative AI models: Will be updated and edited efficiently for safety, effectiveness, and clinical knowledge

• What I Got Right: There’s growing focus on “continual learning” and rapid fine-tuning of models, especially in medical or legal domains needing up-to-date info.
• What Didn’t Come True: Regulators and institutions haven’t fully endorsed generative AI in live clinical settings yet. Concerns about data accuracy, liability, and privacy linger, slowing widespread adoption.

33. AI-based applications: Will be faster, more agile, and more customizable than previous applications

• What I Got Right: Modern AI toolkits, combined with no-code and low-code platforms, allow teams to build and iterate on AI-driven apps at breakneck speed compared to a few years ago.
• What Didn’t Come True: Rapid build times still face roadblocks in heavily regulated sectors or where data privacy is paramount, limiting the extent of customization and deployment.

44. Real-time moderation: Will be enabled on corporate sessions, saving time and money

• What I Got Right: Many enterprise collaboration tools have implemented AI-driven moderation, transcription, summarization, and translation features in real time.
• What Didn’t Come True: True real-time “moderation” is still evolving, especially for sensitive or complex communications. Accuracy and latency issues remain challenges.

35. Work: AI will continue to enhance the way we work

• What I Got Right: AI tools are widely used to automate repetitive tasks, improve decision-making, and foster collaboration across remote or distributed teams.
• What Didn’t Come True: Some predicted a mass displacement of jobs by AI in 2024, which hasn’t fully materialized. Instead, AI often acts as a complementary resource for most knowledge workers.

36. GTM: Gen AI will cut down go-to-market time for enterprises

• What I Got Right: AI-driven prototyping, testing, and even marketing content creation have indeed shortened launch cycles, enabling faster pivots.
• What Didn’t Come True: Large, heavily regulated enterprises still move slower than startups, so while go-to-market times have improved, they haven’t shrunk dramatically in every industry.

37. Forensic AI: Hollywood will invest in companies that detect AI fingerprints to pay themselves and artists

• What I Got Right: There’s a growing interest in watermarking and detection tools for AI-generated media, particularly amid ongoing debates about content ownership.
• What Didn’t Come True: Although there’s attention to digital rights management, we haven’t seen major studio takeovers or huge deals purely for “AI detection” solutions. Partnerships are more common than acquisitions.

38. Chatbots: Big tech will invest in companies like Character.AI that enable users to create human-impersonating chatbots

• What I Got Right: We’ve seen significant investment in AI chatbot platforms, including those that mimic well-known figures or fictional personas.
• What Didn’t Come True: Legal, licensing, and ethical concerns are slowing widespread adoption. High-profile deals exist, but we haven’t hit a saturation point.

39. US Election Deep Fakes: Deep fakes will flood the 2024 elections

• What I Got Right: Concerns about deep fakes spiked, and there have been several notable examples that went viral. Experts, NGOs, and social media platforms are on high alert.
• What Didn’t Come True: “Flood” might be strong—while deep fakes are a threat, many political campaigns are wary of being associated with outright deceptive content that could backfire publicly.

40. 2024 will be about creating partnerships and collaborations with cloud service providers, data storage, and analytical companies

• What I Got Right: Strategic alliances between AI startups and tech giants (AWS, Azure, Google Cloud) have accelerated. This collaboration continues to drive new product launches and custom enterprise solutions.
• What Didn’t Come True: Some partnerships remain more marketing than substance. Though real integrations exist, the full potential of these alliances is still developing.

41. 2024 will be the first full year we see how consumers respond to AI-creative works—both their dollars and emotions

• What I Got Right: Consumers have engaged with AI-generated art, music, and short-form video more than ever, revealing both curiosity and skepticism. Many creative AI projects found niche audiences or viral success.
• What Didn’t Come True: Widespread mainstream acceptance—and willingness to pay large sums—varies. While younger, tech-friendly demographics are open to AI-based creativity, others remain hesitant or uninformed.

2024 AI Highlights: What Actually Happened

• Targeted, Smaller Language Models Organizations embraced domain-specific models that require fewer resources and can be rapidly deployed, making AI more accessible and cost-effective.
• On-Device AI We saw more AI models running locally on phones and specialized hardware, improving privacy and speed while reducing dependency on the cloud.
• Multimodal Explosion Models capable of processing text, images, and audio became mainstream, enabling more immersive and versatile AI applications in healthcare, marketing, and beyond.
• Responsible AI and Governance With new regulations and frameworks gaining traction worldwide, 2024 became a pivotal year for transparency, data ethics, and AI accountability.
• Open-Source Advancements Open-source projects continued to drive innovation, giving developers and enterprises alike the tools to build customized AI solutions at lower cost.
• AI for Accessibility New assistive products leveraged AI for text-to-speech, image recognition, and personalized learning—empowering people with disabilities in daily tasks and employment.
• AI-Generated Video Breakthroughs Tools for AI video generation improved in speed and quality, making it possible to create realistic short clips and marketing content at scale (though still shy of true Hollywood-level VFX).
• Rise of AI Agents Specialized “agents” designed to tackle focused tasks—from coding to sales outreach—began transforming workflows and boosting productivity for teams and individuals.
• Collaboration and Partnerships Tech giants formed new alliances with AI startups, combining cloud infrastructure, data storage, and analytics to accelerate enterprise adoption of AI.
• Real-Time Moderation and Automation Organizations implemented AI-based moderation in virtual meetings, customer support, and collaboration platforms to save time, ensure compliance, and improve user experiences.
• Accelerated Go-to-Market From prototypes to production, AI-driven product development drastically shortened launch timelines, enabling companies to adapt quickly to market demands.

Conclusion While a few predictions (particularly around Apple supplanting Siri or building a direct Google competitor) haven’t fully come to pass, many others—like the emphasis on on-device AI, open source leadership, and the rise of specialized LLMs—were right on the money.

My predictions also captured the broad trajectory of AI in 2024: more specialized models, on-device compute, agent-based workflows, and a rise in open-source ecosystems. Some developments—especially around responsible AI, data strategies, and market disruptions—unfolded, though perhaps not with uniform speed or completeness.

Other forecasts, such as Hollywood-grade AI video and mass adoption of multi-agent “AI Crews,” remain more on the horizon than fully realized. Overall, it was a forward-looking list where I got many things right, and it highlights just how fast this technology continues to evolve.

It’s been an incredible year for AI, and these developments are likely just the tip of the iceberg.

Thanks for reading! What trends have you seen in AI this year, and where do you think it’s headed next?

#AI #GenAI #2024AIPredictions