2023 has seen ChatGPT provide AI with its ‘iPhone moment’, propelling the technology directly into the public consciousness. While the levels of engagement and interest may be new, the underlying technology has been developing for decades. Throughout the recent UBS Private Company Showcase, leading companies demonstrated how all that innovation is driving transformation across nearly every sector of the global economy.
For enterprises, there is no question about the potential of AI to increase efficiencies, boost productivity and enhance decision-making across every function. The uncertainties are now around the practicalities – adoption roadmaps, organizational impact, potential monetization strategies and how best to engage stakeholders and demonstrate the value of AI investments.
We invited leading companies working in sectors like finance and healthcare, along with corporate decision-makers to discuss what has changed, how the market is evolving and how smaller AI companies can Scale and build market share while competing with the likes of Google and Microsoft. Here is what they told us.
Generalist or specialist models?
Generalist or specialist models?
The reason why ChatGPT and Bard have resonated with the public is largely because they function like a WhatsApp or Slack conversation with a colleague. Another reason is that these models are generalists. They are touted as being able to help with almost any task. But being trained on massively broad data sets impacts their ability to produce factual outputs without hallucinations. For highly regulated industries, the risk of generalist models is currently too high.
For enterprises, there is much more value in specialist, analytical AI models. Trained on specialist data, these models are better able to analyze, interpret and create highly industry- or process-specific content. For example, in the pharmaceutical space, specialized AI is already helping to identify target populations for new treatments, standardizing language across drug programs and giving decision-makers a much clearer view of a huge pool of data.
The ‘black box’ risk question
The ‘black box’ risk question
If the future of AI-enabled enterprises involves ecosystems of specialist AI models rather than a generalist model being used for every function, we are likely to see a large universe of AI vendors, model types and service offerings emerge. Enterprises will face the choice between purchasing pre-trained models, training their own or utilizing services based on a pre-trained model.
A ‘black box’ solution may fast track efficiencies, but also brings a lack of transparency for how some AI models arrive at their outputs. This is because with ‘black box’ solution, the internal workings of the AI are by definition invisible to the end user. For highly regulated industries like finance and insurance, this could present a major compliance risk. Businesses in those sectors will have to own their models as well as the risk – and be able to show effective model risk analysis and governance.
Addressing real problems is the key to success
Addressing real problems is the key to success
Across the showcase, one theme came up again and again: how can smaller companies and start-ups compete with tech giants? Whether discussing infrastructure, healthcare, corporate decision making or financial services, several core themes emerged.
In the same way that specialist models can provide better accuracy and relevancy, use cases that address real-life challenges can quickly gain market share. Being able to demonstrate real value creation is a key differentiator. Across the showcase, we saw a number of examples:
- A “doctor in your pocket” app that helps people get instant assessments for symptoms currently has 13 million downloads. Trained by doctors, the app is helping capacity issues at GP offices
- An education initiative aiming to provide students in Africa with a tablet that gives them access to a one-on-one AI tutor that can help them expand their learning
- A solution that designs bus routes, allocates drivers and manages timetables for operators in over 2,000 cities worldwide
On top of being able to demonstrate real value in niche areas, businesses need to show that not only are their models trained on their own data, but that they can scale to meet demand too. Add to that the innovation and agility that smaller companies tend to have and a clear blueprint for competing in the crowded AI market is emerging.
C-suite 2.0
C-suite 2.0
AI is also starting to impact how the C-Suite makes decisions and strategizes for the future. Having instant access to a much wider pool of data provides the opportunity for enterprises to more efficiently identify skills gaps within their organizations, assess risk and drive change across their organizations. CFOs, for example, will be able to much more easily attribute how decision making at every level of a business relates to financial performance.
Augmenting your enterprise
Augmenting your enterprise
The AI revolution isn’t coming – it’s already here. Across the global economy, enterprises are thinking creatively about how to leverage the technology to augment their skilled workforces, make sense of mountains of data and turn insight into action much faster. The evolution of the AI market over the coming years will see specialist models battle for space with the generalist giants. The companies that come out on top will be the ones that can demonstrate real value, data distribution and scale.
