DEV Community

Cover image for AI in 2024: Art Thrives, Open-Source Battles GPT
MindsDB Team for MindsDB

Posted on • Updated on

AI in 2024: Art Thrives, Open-Source Battles GPT

If only there were a crystal ball with a chatbot inside.

*ChatGPT: Tell us what will happen next in AI.
*

Will we all be texting telepathically? Popping popcorn and watching AI-generated movies? (A Marvel movie director says that’ll happen in 2025).

The next best thing to a crystal ball: Polling AI founders, industry experts, and analysts about what they think is in store for 2024.

TL;DR:

For humanity: AI will unleash tremendous creativity but can’t create masterpieces on its own. Early adopters will manage emails with AI; employers will use it to give customer service a boost. For AI practitioners: Open-source AI models and tools will gain ground against GPT. Many AI infrastructure startups, tools, and solutions will emerge to support AI deployment. Companies will integrate AI more strategically integrated with products & services.

Here’s what else our experts see around the corner.

Image description

Navin Chaddha, Managing Partner, Mayfield
Mayfield is a global venture capital firm with $3 billion under management.

  1. Will AI replace knowledge workers in 2024? Nope.

“The future lies in the synergy between humans and AI, enhancing human capabilities through advanced, conversational interactions and cognitive assistance,” Navin Chaddha says. “This collaboration will lead to a concept we refer to as ‘Human Squared,’ where AI acts not just as a tool, but as a teammate, multiplying our own abilities.”

  1. Generative AI’s infrastructure will take a big leap.

“We anticipate significant advancements in startups focusing on gen-AI infrastructure layers, akin to the evolution seen in web, mobile, and cloud technologies,” he says.

Mayfield calls this infrastructure the ‘cognitive plumbing of GenAI,’ and it will consist of four crucial layers:

→ models/middleware/tools

→ data infrastructure/operations

→ infrastructure software/XaaS

→ semiconductors and systems

  1. AI will become integral in applications.

This phenomenon mirrors the necessity of having a website or mobile app today, Chaddha says. The shift is not just about cost-cutting, but about enabling new capabilities previously beyond human reach.

“At the base, we'll see growth in semiconductors and systems, with companies innovating beyond current technologies like Nvidia's GPU and AI processors. The next layer involves infrastructure enhancements, tackling challenges in AI security, networking, and storage. Above this, the data infrastructure and operations layer is critical for running AI effectively. Lastly, the top layer, comprising models, middleware, and developer tools, will see innovation, enabling easier integration and utilization of AI in various services.”

  1. A new crop of founders will tackle complex problems and embark on the long journey of company building.

“These entrepreneurs, dedicated to addressing significant challenges, will find a receptive audience among investors eager to support them from inception to iconic,” says Chaddha. “A focus will be on founders with a deep commitment to their vision and values, coupled with the emotional intelligence to lead and inspire their teams.”

Image description

Eric Buatois, General Partner, Benhamou Global Ventures
BGV is a Palo Alto-based early-stage venture capital firm focused on global, human-centric tech.

  1. Creatives will get a big boost from AI.

Especially those making video games, music, even opera, Eric Buatois says. “The creative people are going to have a blast.” But can AI generate a full-length movie? Not quite. Same for painted masterpieces. “Art is about emotion. I’m not so sure that AI can create an emotion like that yet.”

  1. LLMs will get smaller and more specialized.

“You can’t boil the universe,” Buatois says. “In many cases what’s needed is something that’s more optimized for one field, like an LLM for biochemistry, materials, biology etc.”

  1. LLMs will move from non-essential to mission-critical.

Companies will go far beyond running entertaining experiments in 2024, Buatois says. “As a result, we’ll expect LLMs to be faster and more reliable. You’ll see big improvements in reliability, response time, and cost.”

Image description

Didier Lopes, Founder and CEO, OpenBB
OpenBB is an open-source fintech company that integrates dozens of data vendors into a single fully customizable platform.

  1. AI movies: Not quite yet.

“AI won’t be creating whole hour-and-a-half-long movies in 2024, but it will be creating plenty of backgrounds and scenes. Plus also a lot of short 5-minute films,” says Didier Lopes.

  1. AI for email? Yes, for early adopters.

“Early adopters will have an AI-powered system for managing all their emails,” he says. That includes organizing, sorting, reading, and responding to them. (Although Lopes won't be an early adopter here. “I like to have control and know about each context I have with any individual since that can build a stronger relationship over time.”)

  1. AI for coding: Absolutely.

“AI for coding is not just about tests anymore,” Lopes says. “Look at Cursor. I use it by default now and get the first iteration of code 60-70% done through it, and then through prompt engineering I can squeeze a few other 10-20%. Then I adapt the last 10-20% to my particular use case. This should be the norm, and people who don't adapt are actually losing out on productivity.”

Image description

Richard Socher, CEO, you.com, & Founder, AIX Ventures
Socher is former Chief Scientist and EVP at Salesforce, where he led teams working on fundamental research, applied research, product incubation, search, and other areas.

  1. Open-source AI models will catch up to GPT-4.

“Companies will eventually use LLM operating systems that help you get these (open-source) models production-ready.”

  1. AI-generated videos will get longer.

“They’ll eventually really solve the temporal consistency of multiple characters,” Socher said in a tweet.

  1. LLMs will graduate to becoming agents.

“These agents will have more powerful tools, search, APIs, coding abilities, clicking on the web, etc. In particular, AI assistants for search are helpful enough and will replace Google for many young people, students and knowledge workers.”

  1. AI music generation? Yes.

“The first vibe album may get released where an artist gives a general sequence of vibes. The exact lyrics will be personalized. Before that, we'll have more artists use and collaborate with AI to create new songs.”

Image description

Brendan Burke, Senior Emerging Technology Analyst, PitchBook
Burke contributes to the firm’s emerging technology research, covering AI, information security and IoT.

  1. Open-source AI agent projects = big businesses.

“We are just starting to learn how to use function calls to LLMs to complete complex tasks,” Brendan Burke says. “Experimentation with agents is producing some of the fastest growing open-source projects of all time. These experiments will evolve into reliable applications next year.”

  1. Domain-specific models will take a leap ahead.

“Right now, we are talking about AGI efforts and performance on academic tasks. By next year, individual professions will be comparing new models in their domains from specialized training labs.”

  1. AI directing money flow? Yes.

“AI will absolutely change capital allocation decisions. Banks and investors are at the leading edge of generative AI experimentation and have the ability to dramatically disrupt their investment processes to evaluate more companies with richer context. This may be a way for financial services firms to stand out in performance and innovation.”

Image description

Adam Carrigan, Co-Founder & COO, MindsDB
MindsDB is an end-to-end AI platform for developers. Carrigan is a former management consultant at Deloitte, a University of Cambridge grad, and a YCombinator and UC Berkeley SkyDeck alum.

  1. LLMs will have a big impact on customer service.

“There’s a ton of repetitive work in this area, from help desks to frequently asked questions about products," Adam Carrigan says. "LLMs can really improve the experience here, and there will be quicker and better outcomes for everybody – including end users and companies.”

  1. Open-source LLMs will shine.

“Yes, there’s been a lot of excitement around OpenAI’s models. But in 2024, open-source models will come to the forefront. They’re cheaper, they give companies more control, and using them means you’re no longer beholden to a large organization like OpenAI.”

  1. GPT wrappers are goners.

“A whole crop of startups were built around improving on OpenAI. But if they don’t have a true moat, their businesses stand to get gobbled up by OpenAI itself.”

  1. AI music finally gets good? Maybe.

“There have been some hits this year (“Heart on My Sleeve” by an AI mashup of Drake and The Weeknd) and clear misses (think Anna Indiana). But I’m hopeful that 2024 is the year that we can tune into AI-generated music and actually enjoy it.”

  1. AI hardware: Not yet.

“The Humane AI pin launch was interesting, but 2024 is too soon for AI devices. Early adopters may try out the pin or other gadgets, but this won’t take off yet in a mainstream way.”

Image description

Zoya Khan, CEO and Co-Founder, AfterWork
Khan is a serial entrepreneur, angel investor, and former venture capital analyst.

  1. Human interaction will be fully back, as more people automate repeatable tasks like scheduling calls or appointments.

“People want to get back to interacting with each other and don’t want to be bogged down by all the logistical challenges of planning,” Zoya Khan says. “So I believe AI will be used in creative ways to allow people to connect without the logistical headaches.”

Image description

Divyansh Garg, AI founder and researcher

  1. Skynet family reunion? Garg predicts: “By early 2025, the number of AI agents active at a moment will exceed the population of humans on Earth.”

Image description

Lior Sinclair, AI/ML Engineer and Founder of AlphaSignal
AlphaSignal is one of the best-read technical newsletters for engineers and researchers.

  1. An open-source model will beat GPT-4 in 2024.

  2. LLMs will become smaller. (Much smaller.) Phi2 is a good example.

  3. LLMs will understand math and physics.

Image description

Lindsey Gamble, Associate Director of Influencer Innovation, Mavrck
Gamble was named a Top Creator Economy & Influencer Marketing Expert by Business Insider.

  1. AI will be the key way that creators scale.

“With AI, creators will leverage tools to transform their content into different formats across various social media platforms, such as turning long-form videos on YouTube into multiple short-form videos for YouTube Shorts, Instagram Reels, and TikTok, among others,” Lindsey Gamble says.

  1. AI dubbing tools will take off.

“AI dubbing tools that let creators put out their versions of their videos in different languages will have the most impact by allowing them to tap into audiences they previously couldn’t reach due to the language barrier.

This is something that only the biggest creators had the resources to do previously, but with YouTube and other standalone companies building out tools like this, it will create more of an equal playing field for creators to build and grow their audiences and monetize them while reaching fans across the globe.”

  1. Get ready for a wave of digital doppelgangers.

“More creators will create AI versions of themselves, ranging from chatbots to avatars, built on their existing content,” says Gamble. “Chatbots will be the main entry point for creators, especially B2B creators.”

Image description

Tim Ruscica, Developer, Educator, and Influencer
Ruscica's @TechWithTim YouTube channel has more than 1.39 million subscribers.

  1. AI won’t replace programmers.

“Yes, AI is advancing very quickly, and surely it will affect the productivity and demand for conventional developers. However, I don't see it replacing them,” Tim Ruscica says.

Companies still need developers, they need people that have the knowledge to understand, put in context and implement solutions provided by AI. AI is already changing how developers work but those developers are the ones that can use it most effectively and gain the most benefit from it.”

  1. AI-assisted writing will go mainstream.

“People have relied on auto-complete and spell check for a very long time already, and now AI is great at augmenting writing, especially with extensions and plugins for things like email. I think it will be more common for people to simply rely on it for almost all of their writing.”

I can see a world where professionals merely list a few points in notes and have AI fill in the rest. (Leading many of us to wonder whether we are reading or responding to AI-generated content.) I think this will vastly reduce the literacy barrier for many people, especially in nations where English is not their first language.”

  1. We’ll develop a new appreciation for art.

“At the same time, I do believe that with more AI-generated content around, we will slowly learn to appreciate other forms of art, music and creativity produced by humans. Many people will resist reliance on technology for creative pursuits and strive more and more to do things ‘naturally’.”

Image description

Kirk Borne, Founder, Data Leadership Group
Borne is a data scientist, influencer, speaker, and consultant.

  1. 2024 should be about the “why” of AI.

“As incredible as large language models (LLMs) have been in 2023, the focus on ‘what’ these are will diminish as the focus increases on the more strategic business questions: ‘why’ and ‘where’ should we be using these AI developments within our organization, in a manner consistent with our business goals, mission, culture, and go-to-market strategies."

  1. AI will be more strategically integrated.

“I believe we will see more AI enablement within existing business tools and processes, with less concern about figuring out how to do tactical stand-alone deployments of the latest, greatest hyped-up AI tools.

I also believe we will see greater appreciation of data strategy (data quality, metadata labeling, data integration, data workflow orchestration, data-enablement of AI deployments) within the realm of AI implementations, since AI not only consumes data, but AI devours data.

Consequently, good data in, good AI results out — versus the undesirable alternative.”

  1. Personal LLMs are what’s next.

“We first saw LLMs tackle worldwide content on the Internet. We next saw LLMs focus on on-prem private enterprise knowledge bases and data sources.

I now want to see my own personal, privacy-protecting, smart LLM deployed on my own data sources (email, personal computer, My Documents, search histories, Internet favorites, workflows, data handling tasks, etc.)”

Top comments (0)