This last year was a boom for artificial intelligence, as the technology went from niche to mainstream about as quickly as anything else. 2024, however, will be the year the hype becomes reality, as people consider the potential and limitations of artificial intelligence in general. Here are a few ways we think it will.
OpenAI is becoming a product company
After the leadership reshuffle in November, OpenAI will be a changed company — maybe not externally, but the fallout from Sam Altman in charge will be felt at every level. And one of the ways we expect it to manifest is in the “send it” mentality.
We’ll see this with the GPT store, which was originally scheduled to launch in December, but understandably delayed due to C-suite spasms. The “app store for AI” will be pushed hard as The platform to get your AI games and tools from Hugging Face or any open source model. They have a great model to work with, Apple’s, and they’re going to follow it all the way to the bank.
Expect more such moves from OpenAI in 2024, as the caution and academic reserve exercised by the previous board gives way to an inappropriate lust for markets and customers.
Other big companies with AI efforts will also follow this trend (for example, expect Gemini/Bard to attract a ton of Google products), but I suspect it will be more pronounced in this case.
Agents, video creation and music production graduate from visual to experimental
Some niche applications of AI models will grow beyond the “eh” state in 2024, including agent-based models and genetic multimedia.
If AI is going to help you do more than summarize or create lists of things, it will need access to things like your spreadsheets, ticketing interfaces, transportation apps, and so on. 2023 saw a few tentative attempts at this “agent” approach, but none really caught on. We don’t really expect any to take off in 2024, but dealer-based models will show their stuff a little more convincingly than they did last year, and some clutch use cases will emerge for notoriously tedious processes like making insurance claims.
Video and audio will also find places where their disadvantages are not so visible. In the hands of experienced creators, the lack of photorealism isn’t a problem, and we’ll see video AI used in fun and interesting ways. Likewise, music production models will likely become some major productions, such as games, where professional musicians can leverage the tools to create an endless soundtrack.
The boundaries of monolithic LLMs are becoming clearer
So far there has been great optimism about the capabilities of large language models, which have indeed proven more capable than expected, and have grown correspondingly more as more computations are added. But 2024 will be the year that something gives. Exactly where is impossible to predict, as research operates at the frontier of this field.
The seemingly magical “emerging” capabilities of LLMs will be better studied and understood in 2024, and things like their inability to multiply large numbers will make more sense.
At the same time, we will start to see diminishing returns in parameter measurements, to the point where training a model of 500 billion parameters can technically produce better results, but the computation required to do this could demonstrably be deployed more efficiently. A single monolithic model is unwieldy and expensive, while a mixture of experts—a collection of smaller, more specialized models and possibly multimodal models—can prove nearly as effective while being much easier to update piecemeal.
Marketing meets reality
The simple fact is that the hype created in 2023 will be very difficult for companies to sustain. The marketing claims made about machine learning systems that companies have adopted to keep up are going to get their quarterly and annual reviews… and they are very likely to be found wanting.
Expect significant customer churn from AI tools as the benefits don’t justify the costs and risks. At the far end of this spectrum, we are likely to see lawsuits and regulatory action against AI service providers who have failed to back up their claims.
While the capabilities will continue to grow and evolve, the products of 2023 will not all survive by a long shot, and there will be a round of consolidation as wavering wave riders fall and are consumed.
Apple steps in
Apple has a well-established pattern of waiting, watching, and learning from other companies’ failures, and then blowing it with a refined and refined point of view that puts others to shame. The time is right for Apple to do this in AI, not only because if it waits too long, its competition could destroy the market, but because the technology is ripe for its improvement.
I’d expect an AI that focuses on practical applications of user data, using Apple’s increasingly central place in their lives to integrate the many brands and ecosystems the company is familiar with. There will probably also be a smart and elegant way to handle problematic or dangerous messages, and while it will almost certainly have multimodal understanding (mostly for handling user images), I imagine they’ll skip media generation altogether. Expect some narrowly tailored but impressive agent capabilities, too: “Siri, get a table for 4 at a downtown sushi place around 7 and book a car to pick us up.”
What’s hard to say is whether they’ll bill it as an improved Siri or an entirely new service, Apple AI, with a name you can choose yourself. They may feel the old brand is saddled with years of comparative incompetence, but millions already say “hey Siri” every 10 seconds, so they’re more likely to choose to keep that momentum going.
Legal cases are built and broken
We saw a fair number of lawsuits filed in 2023, but few saw any real movement, let alone success. Most lawsuits over copyright and other wrongdoing in the AI industry are still pending. 2024 will see many of these fall to the wayside as companies withhold critical information such as data and training methods, making claims such as the use of thousands of copyrighted books difficult to prove in court.
That was only the beginning, however, and many of these lawsuits were filed essentially on the beginning. While they may not succeed, they may open up the process enough during testimony and discovery that companies would rather settle than have certain information come to light. 2024 will also bring new lawsuits, those related to the misuse and abuse of AI, such as wrongful termination, bias in hiring and lending, and other areas where AI is put to work without much thought.
But while some egregious examples of misuse will be punished, the lack of relevant laws specifically addressing this means that it will necessarily only be brought to court by accident. On that note…
Early adopters of the new rules follow the horns
Big moves like the EU’s AI law could change the way the industry operates, but they tend to be slow to come into force. This is by design, so companies don’t have to adapt to new rules overnight, but it also means we won’t see the impact of these big laws for quite some time, except for those willing to make changes proactively and voluntarily. There will be a lot of “we’re starting the process of…” conversations. (Also expect some quiet lawsuits challenging various parts of the laws.)
To that end, we can expect a new booming AI compliance industry, as the billions poured into the technology require matching investments (on a smaller scale, but still significant) to ensure tools and processes meet international and local standards.
Unfortunately for anyone hoping for meaningful federal regulation in the US, 2024 is not the year to expect movement on this front. While it will be a year for artificial intelligence and everyone will be calling for new laws, the US government and electorate will be too busy with the dumpster fire that will be the 2024 election.
The 2024 election is a garbage fire and AI is making it worse
How the 2024 presidential election will play out is, really, anyone’s guess at this point. There’s too much up in the air to make any real predictions, except that, as before, influencers will use every tool in the box to move the needle, including artificial intelligence in whatever form is convenient.
For example, expect bot accounts and fake blogs to spout nonsense 24/7. Some people who work full-time with a text and image generator can cover a lot of space, creating hundreds of social media and blog posts with completely made-up images and news. “Flooding the zone” has always been an effective tactic, and now AI acts as a job multiplier, allowing for more massive yet targeted campaigns. Expect both false positives and false negatives in a concerted effort to confuse the narrative and make people distrust what they see and read. This is a profitable state for those politicians who thrive on chaos.
The organizations will tout “artificial intelligence” analytics to support voter roll purges, vote count challenges and other efforts to suppress or interfere with existing processes.
The resulting video and audio will join the fray, and while neither is perfect, they’re good enough to be believable with a bit of blur: The clip doesn’t need to be perfect, because it’ll come across as a grainy zoomed-in mobile shot phone in a dark room or a hot mic at a private event or what have you. Then it becomes a matter of “who are you going to believe, me or him?” And that’s all some people need.
There will probably be some half-hearted attempts to block the use of generated content in this way, but these posts can’t be removed fast enough by the likes of Meta and Google and the idea that X can (or will) effectively monitoring and removal of such content is unlikely. It’s going to be a bad time!
