What's Next for Generative AI?
Artificial intelligence has enormously impacted how people acquire information. With generative AI, those capacities also include how people create and share information. The possibilities are almost wholly unlimited. But learning from generative AI's major early applications will prove essential to wielding it more effectively with every iteration.
Continue reading to explore the current state of generative AI, ongoing questions, and what's to come.
Who's Driving Generative AI Forward? A Marketing View
According to Dentsu's research, no less than 91% of CMOs believe generative AI marks "the future" of digital marketing. Given that 78% of consumers agree, these sentiments indicate massive preemptive buy-in from advertisers and their audiences.
Marketing departments for global consumer brands have already begun large-scale applications of generative AI content. What do these early efforts teach us about the value of generative AI and which applications it's currently best suited to?
New Developments in 2024 and Beyond
The most ambitious generative AI marketing firms have built teams around its use, bolstered with skills development programs that major tech companies endorse. It isn't hard to see why. Considering content production capabilities alone, generative AI breaks innumerable barriers, particularly reduced costs, faster production, and fewer limits on content.
Business-wise, the same advertisers (mostly firms for large household brands) also wield the most significant market data repositories. We've seen how such companies generally apply market data to content strategies. But it's anyone's guess how they'll need to adjust their approach to best capitalize on generative AI content.
Changing What It Means to Search
The most uncharted terrain for generative AI will likely be a byproduct of its effects on web searching. Current search engine functions took about a quarter-century of effort. However, competing for web traffic now compels search engines to leverage generative AI content.
The reason is less related to the content itself and more to the habits of advertisers and consumers. As companies successfully populate search results with generative AI ad copy, they have less incentive to fuel Google's primary commercial interests by purchasing ad space.
Users also have less incentive to meet brands on search-result pages. Why would they, if they can outsource those functions to their preferred artificial intelligence assistant? Consider it a choice between two fundamentally different approaches:
Begin your search with a list of search engine results.
Essentially, end your search using artificial intelligence.
A New Continuum in Language Development
These developments are only possible because artificial intelligence has made tremendous technical gains in natural language processing (NLP) and large language models (LLMs). It's made language-driven human-machine interfaces (HMIs) much more valuable and usable, increasing its reach to more users.
Now, anyone can mine massive data repositories (about the size of a novel) almost as quickly as they once explored data on a single small web page. With new vistas of processing power and more sophisticated contextual queues, artificial intelligence is about to atomize, then reassemble, the web searching process.
It would seem how we "talk" to machines is about to change—and with natural language generation (NLG) adding to that lexicon daily, it's also changing how machines talk back.
Generative AI's Strengths and Weaknesses: Watch This Space
As companies increasingly invest resources in generative AI, the need to guess its efficacy will diminish accordingly.
Shopify Magic, for example, recently gave users the ability to auto-generate blog copy and product descriptions (but will it move products?). Web hosting companies increasingly use generative AI as a selling point (does it increase or diminish web traffic?). Salesforce has devoted significant development resources to products built on OpenAI (will it outperform the alternatives?).
Soon enough, every industry will be inundated with cold, hard analytics showing the efficacy of generative AI products and services.
More telling are the budding technical capacities of artificial intelligence. For instance, GPT-4 boasted a 40% improvement over GPT -3.5. In just two months, Anthropic's Claude surpassed its language processing limits by tenfold.
Being an integral new part of Microsoft Office 365, ChatGPT now has a steady dose of performance data to mine. As generative AI tools attract more users, they secure future performance upgrades. In this crucial way, the artificial neural networks AI is based on have proven more like than unlike the real thing.
The Most Suitable Applications of Generative AI
As described, enterprise-scale advertisers are initiating the bulk of generative AI activity now permeating global economies and populations. What about small and mid-market businesses, though — which value-adding activity is generative AI best suited for, overall? The answer holds the key to unlocking the most significant shares of the estimated $2.6–4.4 trillion from generative AI each year.
McKinsey pared generative AI's most valuable functions into four distinct categories:
Customer operations. Automated service prompts and responses will provide timely and relevant support and value-adding touchpoints to end users.
Marketing and sales. The ability to quickly create new content based on a growing number of parameters will lead to greater personalization and targeted messaging.
Software engineering. As people increasingly expect polished digital services from the companies they support, there's pressure to develop better digital tools at lower costs.
Research and development. Sheer processing power will prove indispensable to applying large swaths of market and product research most effectively.
As for which industries benefit most, McKinsey considers all industry sectors subject to the effects of generative AI. However, they singled out banking and high-tech as the industries with the most to gain—several hundred billion in each case.
Disruptions and All
The same McKinsey research contemplates that improved NLP will drive greater labor automation because a quarter of all work depends on natural language. That's happening faster than anticipated, giving reason to expect faster replacement of "knowledge work" (fields more dependent on higher education).
As AI becomes more valuable to specialized industries, their labor pools increasingly ask the hard questions previously reserved for unskilled labor markets. Workforces at large may be facing more similar than dissimilar positions.
There are also critical unanswered questions about generative AI copyright claims. These include:
When is creative digital output the product of an individual vs. a string of code?
To what extent should authors of generative AI apps receive creator's rights?
Do products of open-source generative AI become public domain?
Even that line of questioning assumes complete certainty over when artificial intelligence takes or mistakes copywritten work. Some are raising broader ethical questions about the potential for generative AI content to harm or deceive people.
To these ends, some have tried imbuing artificial intelligence with a "constitution" based on "human principles." At least, that's how Anthropic has framed efforts to ensure their artificial intelligence is "helpful, harmless, and honest."
A question yet to be answered is how the most ethical generative AI would fare in the marketplace of ideas.
As technology develops at breakneck speed, Think Global Forum continues to foster thought-provoking conversations about artificial intelligence and its broader implications. Join the Think Global Forum community to receive more insights. To dive deeper into the impact of AI, catch up on our latest Think Global Forum AI Roundtable. The key learnings are available here.