Navigating the evolution of generative AI: what’s next for organizations

Organizations (and society) must balance AI innovation with issues of value, scalability, ethics and sustainability.

In brief

• The evolution of generative AI is exposing gaps in value realization, governance and sustainability.
• Organizations must shift from siloed experimentation to AI-first process reinvention.
• Proactive regulation, ethical alignment, and sustainable innovation are key to long-term success.

GenAI exploded into the mainstream two years ago, rewriting the rules of life and work as we know it. Now, as the depth and breadth of the technology’s integration grows, there are signs it’s experiencing a period of growing pains akin to the “terrible twos.”

Organizations (and society) must balance AI innovation with issues of value, scalability, ethics and sustainability.

Caught in the innovation loop

As organizations strive to grow their AI usage, many are also being challenged by the rapid evolution of generative AI. This can lead to an analysis paralysis in which decision-makers become caught in a loop of trying to identify the right solution for their business, only to find it almost immediately surpassed by a new model. They then set about analyzing this new solution rather than actually investing.

The problem is that waiting for innovation to calm down can leave organizations behind competitors. Far better, then, to focus on building a team — through both training, recruitment and partnerships — that’s flexible and dynamic enough to use the newest type of model or tech and then evolve with it as use cases and technology change.

There are also architectural considerations to be taken into account. AI systems need to have a modular approach that enables the ability to swap out some components as better options rapidly become available.

Taking responsibility

Responsible development and deployment are also complex and pressing issues, requiring firms to demonstrate clear and effective action around bias, data governance, privacy and environmental impact. This includes investing in areas like red and purple team testing to assess their security posture, model effectiveness, and independent validation of ethical outcomes.

Also under consideration should be ways to limit or offset the carbon footprint of their AI systems. Crucially, companies must have an “AI ethical compass” that helps guide them to have GenAI use cases that align with their own and their customers’ ethics and values.

And finally, there’s the issue of surging demand for graphics processing units (GPUs), the backbone of GenAI models. The more these are used, the scarcer and more expensive they become. And because they consume a lot of energy, they present a challenge from a sustainability perspective too. Innovation in this space is happening rapidly, but it will take time to catch up to the volume of usage and ability to be leveraged at scale.

Addressing cost and sustainability concerns

According to the December 2024 EY AI Pulse Survey of 500 senior business leaders, “many expressed concern about the downsides of increased AI usage, specifically the cost implications (69%), the negative impact on their sustainability and emissions goals (64%), and the reliability of their energy supply due to increased AI use (62%).”

Exploring more sustainable alternatives will be vital. These include:

• MIN AI models, which operate with limited input and training data
• Intelligence scaling, which aims to increase AI models’ algorithmic capabilities without growing the amount of data required
• Agentic AI, which can autonomously carry out tasks across entire workflows

We’re even seeing the emergence of innovations like ultra-compressed memristor-based storage to reduce the amount of resources AI and data solutions consume. All of these offer exciting potential in letting organizations deliver significant operational improvements and efficiencies without compromising their sustainability goals.

Proactive, not reactive

As you might expect, the solution to AI’s terrible twos doesn’t lie with a single silver bullet, nor will the path to full adoption be uniform. Some industries, particularly highly regulated ones like financial services and life sciences, will also move faster than others due to being accustomed to significant regulatory scrutiny and having therefore created robust data governance structures to match.

Yet something all leaders can do to accelerate their organization’s GenAI transformation is take a proactive stance, not a reactive one, particularly when it comes to regulation. This means engaging with state and federal governments to co-create a legislative landscape that balances incentivization and control and is far more nuanced than simply copying and pasting laws from Europe.

Like everything around GenAI, this partnership between the public and private sectors is ongoing. As the technology matures, the cultural, economic and technological shifts it creates will become more significant every year. The key for every organization is to keep pace now — not wait and try to catch up later.

By focusing on holistic process transformation, investing in sustainability, and acting on responsible governance, leaders can navigate the AI growing pains of today and position themselves for even greater success tomorrow.

This article was originally published on FastCompany.com.