With 2023 in the rearview mirror, it’s fair to say that OpenAI’s release of ChatGPT just over a year ago threw the tech industry into an excited, manic state. Companies like Microsoft and Google have thrown tremendous resources at AI in order to try to catch up, and VCs have tripped all over themselves to fund companies doing the same. With such a tremendous pace of innovation, it can be difficult to spot what’s coming next, but we can try to take clues from AI’s evolution so far to predict where it’s headed. Here, we present 10 bold predictions laying out how emerging trends in AI development are likely to play out in 2024 and beyond.
1. Personal AI Trained on Your Data Becomes the Next Big Thing
While some consumers were awed by the introduction of ChatGPT, perhaps many more picked it up, played with it, and moved on with their lives. But in 2024, the former audience is likely to re-engage with the technology, as the trend towards personal AI will revolutionize user interactions with technology. These AI systems, trained on individual user data, offer highly personalized experiences and insights. For example, Google Gemini now integrates with users’ Google Workspace data, enabling it to leverage everything it knows about their calendars, documents, location, chats and more. Meanwhile, companies like Apple and Samsung are likely to emphasize on-device AI as a key feature, prioritizing privacy and immediacy. It’s not hard to imagine a personal AI with access to all of your data acting as a relationship, education, and career coach, becoming a more integral, personalized part of everyday life.
2. AI Democratizes Data for Business Users (Finally!)
Long promised, the dream of “data democratization” will finally come to pass. AI is finally democratizing data access for business users by enabling them to ask data questions in plain English, eliminating the need to write queries in SQL. Cloud data platforms like Snowflake and Databricks are already falling all over themselves to integrate these features into the next generation of their product offerings.
3. Data Integration Tools Become Key Accelerators in the Race Toward AI
As enterprises focus on centralizing data from diverse operational data stores in order to build a corpus to train AI on, data integration tools are becoming increasingly crucial. The emphasis is on tools that are user-friendly and capable of moving large volumes of data, while keeping that data in sync and up-to-date through the use of change data capture capabilities. As companies race towards AI, the need to leverage these tools to bypass large amounts of technical complexity and beat competitors to market will continue to grow in importance.
4. ‘Governance, Governance, Governance’ Becomes a Mantra for Organizations Serious About Achieving AI
Data governance remains a red-hot topic in the enterprise data management space. Companies at the vanguard of this trend are establishing robust governance frameworks to ensure data quality, compliance and security. This trend reflects the growing awareness and importance of responsible data management practices in a data-rich world. Data catalogs, metadata tagging tools, and data quality tools will leverage AI to great effect, allowing companies to make sense of their data in a more organized, automated fashion.
5. AI Safety and Security Remain a Key Focus for Managing Reputational Risk
Enterprises are increasingly conscious of AI safety and security, particularly in safeguarding brand reputation. Larger companies, having learned from early missteps like the Tay chatbot and biased recruiting tools, are increasingly keen to implement robust guardrails around publicly deployed AI. Their main focus will be balancing innovation with responsibility, aiming to avoid public missteps and ensure fairness and unbiased outcomes. Expect companies to be overly cautious at first, and fine-tune the balance as they become more comfortable.
6. Domain-Specific, Specialized Models Rule the Enterprise
Enterprises are increasingly favoring domain-specific, specialized AI models over general-purpose ones. While an LLM like ChatGPT is quite good at a broad range of general tasks, from writing poetry to summarizing emails, this is less interesting and useful for enterprises than specialized GPTs trained on focused datasets that are excellent in a single domain. For example, a healthcare company might train a GPT on its vast amounts of historical billing data, with the sole purpose of predicting costs with great precision. These tailored models offer greater accuracy and efficiency in enterprise applications, signifying a move towards more customized AI solutions in the corporate world.
7. Open-Source Models Close the Gap
The landscape of AI is witnessing a significant shift as open source models from Mistral, Anthropic, MosaicML and others rapidly improve, narrowing the gap with commercial counterparts like OpenAI. This trend is reshaping the AI ecosystem, making advanced AI tools more accessible and fostering a more competitive and diverse AI market. The pace of innovation seems to only be accelerating, as open source and commercial AIs alike look to come out on top in the AI gold rush.
8. AI’s ‘Wild West’ Era Gets a Little Tamer as Regulation and Compliance Take Hold
Governments and businesses are increasingly focused on regulating AI due to growing concerns about its risks. Governments want to get ahead of the possibility of rogue AIs falling into the hands of bad actors, which could and threaten national security. Meanwhile, some industry players have signaled a desire to expand their moat by supporting rules that limit competition, while others are genuinely concerned about AI’s potential to harm humanity. All of these groups see reasons to act, and as a result, new regulations are starting to take shape. The European Union has already set a precedent with landmark AI regulation that could serve as a model for the USA and others. What that final legislation looks like, though, remains to be seen.
9. Data Lakes Continue to Gain in Popularity
Data lakes are growing at a rapid pace, and finally being taken seriously by large enterprises who are recognizing that they are needed to house large volumes of unstructured and semi-structured text data needed for AI. Data warehouses will still command the majority of market share, but data lakes will continue to grow at a much faster rate. The flexibility and scalability of data lakes make them increasingly attractive for managing large, varied datasets in modern data ecosystems, like those that companies are looking to bring together to train LLMs.
10. Fine-Tuning Models Becomes Significantly Easier
The process of fine-tuning AI models is becoming significantly easier, thanks to new AI platforms that promise a more user-friendly and refined experience. These new platforms will abstract away much of the complexity of fine-tuning, making model customization more accessible, and allowing a broader range of users to tailor AI models to specific needs and applications.
A Final Thought
In the last few waves of highly hyped tech trends (I’m looking at you, crypto), the most prominent players in the space produced a similar amount of sound and fury. Critics argue that AI is no different, and that today’s buzz will settle down once we realize that we need more advanced models to achieve artificial general intelligence (AGI). However, I believe AI is different. Unlike crypto, AI’s practical benefits are already clear to most users, and yet we’re only beginning to grasp how it will be used in innovative and transformative ways. The Age of AI is here, and my hope is that with these predictions, you’ll begin to see where we’re headed, and how profoundly I believe AI will alter the course of industry and humanity alike.
With 2023 in the rearview mirror, it’s fair to say that OpenAI’s release of ChatGPT just over a year ago threw the tech industry into an excited, manic state. Companies like Microsoft and Google have thrown tremendous resources at AI in order to try to catch up, and VCs have tripped all over themselves to fund companies doing the same. With such a tremendous pace of innovation, it can be difficult to spot what’s coming next, but we can try to take clues from AI’s evolution so far to predict where it’s headed. Here, we present 10 bold predictions laying out how emerging trends in AI development are likely to play out in 2024 and beyond.
1. Personal AI Trained on Your Data Becomes the Next Big Thing
While some consumers were awed by the introduction of ChatGPT, perhaps many more picked it up, played with it, and moved on with their lives. But in 2024, the former audience is likely to re-engage with the technology, as the trend towards personal AI will revolutionize user interactions with technology. These AI systems, trained on individual user data, offer highly personalized experiences and insights. For example, Google Gemini now integrates with users’ Google Workspace data, enabling it to leverage everything it knows about their calendars, documents, location, chats and more. Meanwhile, companies like Apple and Samsung are likely to emphasize on-device AI as a key feature, prioritizing privacy and immediacy. It’s not hard to imagine a personal AI with access to all of your data acting as a relationship, education, and career coach, becoming a more integral, personalized part of everyday life.
2. AI Democratizes Data for Business Users (Finally!)
Long promised, the dream of “data democratization” will finally come to pass. AI is finally democratizing data access for business users by enabling them to ask data questions in plain English, eliminating the need to write queries in SQL. Cloud data platforms like Snowflake and Databricks are already falling all over themselves to integrate these features into the next generation of their product offerings.
3. Data Integration Tools Become Key Accelerators in the Race Toward AI
As enterprises focus on centralizing data from diverse operational data stores in order to build a corpus to train AI on, data integration tools are becoming increasingly crucial. The emphasis is on tools that are user-friendly and capable of moving large volumes of data, while keeping that data in sync and up-to-date through the use of change data capture capabilities. As companies race towards AI, the need to leverage these tools to bypass large amounts of technical complexity and beat competitors to market will continue to grow in importance.
4. ‘Governance, Governance, Governance’ Becomes a Mantra for Organizations Serious About Achieving AI
Data governance remains a red-hot topic in the enterprise data management space. Companies at the vanguard of this trend are establishing robust governance frameworks to ensure data quality, compliance and security. This trend reflects the growing awareness and importance of responsible data management practices in a data-rich world. Data catalogs, metadata tagging tools, and data quality tools will leverage AI to great effect, allowing companies to make sense of their data in a more organized, automated fashion.
5. AI Safety and Security Remain a Key Focus for Managing Reputational Risk
Enterprises are increasingly conscious of AI safety and security, particularly in safeguarding brand reputation. Larger companies, having learned from early missteps like the Tay chatbot and biased recruiting tools, are increasingly keen to implement robust guardrails around publicly deployed AI. Their main focus will be balancing innovation with responsibility, aiming to avoid public missteps and ensure fairness and unbiased outcomes. Expect companies to be overly cautious at first, and fine-tune the balance as they become more comfortable.
6. Domain-Specific, Specialized Models Rule the Enterprise
Enterprises are increasingly favoring domain-specific, specialized AI models over general-purpose ones. While an LLM like ChatGPT is quite good at a broad range of general tasks, from writing poetry to summarizing emails, this is less interesting and useful for enterprises than specialized GPTs trained on focused datasets that are excellent in a single domain. For example, a healthcare company might train a GPT on its vast amounts of historical billing data, with the sole purpose of predicting costs with great precision. These tailored models offer greater accuracy and efficiency in enterprise applications, signifying a move towards more customized AI solutions in the corporate world.
7. Open-Source Models Close the Gap
The landscape of AI is witnessing a significant shift as open source models from Mistral, Anthropic, MosaicML and others rapidly improve, narrowing the gap with commercial counterparts like OpenAI. This trend is reshaping the AI ecosystem, making advanced AI tools more accessible and fostering a more competitive and diverse AI market. The pace of innovation seems to only be accelerating, as open source and commercial AIs alike look to come out on top in the AI gold rush.
8. AI’s ‘Wild West’ Era Gets a Little Tamer as Regulation and Compliance Take Hold
Governments and businesses are increasingly focused on regulating AI due to growing concerns about its risks. Governments want to get ahead of the possibility of rogue AIs falling into the hands of bad actors, which could and threaten national security. Meanwhile, some industry players have signaled a desire to expand their moat by supporting rules that limit competition, while others are genuinely concerned about AI’s potential to harm humanity. All of these groups see reasons to act, and as a result, new regulations are starting to take shape. The European Union has already set a precedent with landmark AI regulation that could serve as a model for the USA and others. What that final legislation looks like, though, remains to be seen.
9. Data Lakes Continue to Gain in Popularity
Data lakes are growing at a rapid pace, and finally being taken seriously by large enterprises who are recognizing that they are needed to house large volumes of unstructured and semi-structured text data needed for AI. Data warehouses will still command the majority of market share, but data lakes will continue to grow at a much faster rate. The flexibility and scalability of data lakes make them increasingly attractive for managing large, varied datasets in modern data ecosystems, like those that companies are looking to bring together to train LLMs.
10. Fine-Tuning Models Becomes Significantly Easier
The process of fine-tuning AI models is becoming significantly easier, thanks to new AI platforms that promise a more user-friendly and refined experience. These new platforms will abstract away much of the complexity of fine-tuning, making model customization more accessible, and allowing a broader range of users to tailor AI models to specific needs and applications.
A Final Thought
In the last few waves of highly hyped tech trends (I’m looking at you, crypto), the most prominent players in the space produced a similar amount of sound and fury. Critics argue that AI is no different, and that today’s buzz will settle down once we realize that we need more advanced models to achieve artificial general intelligence (AGI). However, I believe AI is different. Unlike crypto, AI’s practical benefits are already clear to most users, and yet we’re only beginning to grasp how it will be used in innovative and transformative ways. The Age of AI is here, and my hope is that with these predictions, you’ll begin to see where we’re headed, and how profoundly I believe AI will alter the course of industry and humanity alike.