The year 2023 ended on a high note for Artificial Intelligence. OpenAI and its large language model, ChatGPT, spread like wildfire across newsrooms and corporate boardrooms. By December, its public API boasted over 200 million presumed active users.
However, existential doubts about the future of this technology began to surface. Critics pointed to the limited professional adoption, with over 90% of businesses actively banning OpenAI’s tools by the end of 2023.
Then came the “hallucinations” – the buzzword of 2023 – that made headlines. Indeed, ChatGPT boldly claimed that cow eggs were blue, unlike chicken eggs, undermining its potential for enterprise deployment. Worse still, OpenAI, backed by Microsoft, risked creating a monopolistic ecosystem in technology by going it alone.
Adding fuel to the fire, widespread concerns about risks led to a high-profile call to pause AI research, signed by hundreds of prominent figures.
Meanwhile, Europe’s much-anticipated AI regulation, meant to provide reassurance by framing clear usage guidelines, was slow to materialize.
2024 Proved the Critics Wrong!
When it came to practical adoption, 2024 turned the tide. By the end of the year, over 90% of businesses actively encouraged the use of generative AI, with 29% having already trained more than a quarter of their workforce. The cherry on top? Fifty-eight percent of regular enterprise users reported saving at least five hours per week – and not just any hours, but the most tedious ones!
For employees, this newfound time was a boon, allowing them to focus on preferred tasks, strengthen social bonds with colleagues, or spend more time with family.
The much-feared “hallucination” crisis fizzled out. Why? Businesses harnessed large models for their strengths in interaction, analysis, and synthesis. Crucially, the data used to generate responses was proprietary and carefully curated.
Today, AI-driven solutions engage with millions of customers, generate RFPs, and create compliance documents.
The secret? A rigorous protocol of testing, calibration, and experimentation before any large-scale deployment.
Competition Heats Up
The competitive landscape underwent a seismic shift. OpenAI – and by extension, the United States – no longer had the field to themselves. Other American tech giants rolled out their own models: Google’s Gemini, Meta’s Llama, and X’s Grok. Startups also made waves, such as the U.S.-based Anthropic and France’s Mistral, which achieved a staggering valuation of €6 billion just 18 months after launch – a record-breaker!
The Global South wasn’t far behind, with players like 01.AI in China, ALLaM in the Middle East, and Bharat in India. Today, both commercial and open-source solutions offer a wealth of alternatives.
Regulation and Research Progress
The much-debated pause on generative AI research never materialized, and regulatory frameworks began to take shape. On May 21, 2024, the European Parliament published the EU AI Act, aiming to balance the need to protect individuals with the imperative to foster innovation. While businesses remained wary of the current text, they recognized the importance of engaging in the second phase of the legislative process: sector-specific regulations.
The degree to which companies allocate resources to co-create these regulations with policymakers will shape the final outcomes.
2025: Brimming with Potential
Although only 24% of businesses currently report significant gains from AI, it ranks among the top three strategic priorities for 74%. Clearly, companies believe in AI!
Innovation continues, with a new generation of models capable of solving complex problems by breaking them into smaller sub-problems.
Additionally, advancements are facilitating the integration of generative AI into robotics.
And let’s not forget the buzzword of 2025: “agents.” These autonomous systems will encapsulate the power of generative AI, adapting to complex scenarios. In fact, transformers – the algorithms underpinning today’s large models – are on the verge of being replaced by even more efficient successors.
No Clouds on the Horizon? Not Quite.
There are, of course, challenges: model costs, cybersecurity, and malicious uses – the usual suspects for any emerging technology.
But AI has a unique shadow: its insatiable energy demand. The surge in usage raises fears of an unsustainable spike in energy consumption.
The industry is responding with innovations in energy-efficient chips, data centers, and leaner models. The race between energy efficiency and rising demand is on, and its resolution will depend on innovation, pricing, or regulation.
