AI Data Visualization
2024 will likely be the big year to prove the value of generative AI. Some may think that generative AI is a new thing, but it really isn’t, as generative AI was used in the 1960s in chatbots. But the big buzz started in 2014 when generative adversarial networks, often called GANs, were invented. It is a machine learning algorithm capable of creating precise images of videos and audio which has accelerated this technology.
The generative AI bubble has burst with all the tech titans waking up to the generative AI fanfare late last year and now the consulting market frenzy. I can’t think of a single large consulting firm that hasn’t put AI at the forefront of its service offerings. If this is not the case, they are crying out to understand it and wisely.
Deloitte recently released an insightful research report on generative AI, adding rich market research and educational content to fuel more meaning. What’s interesting about reading the report is that they applied some emotional sentiment to what their research respondents felt about generative AI and Eureka – enthusiasm was through the roof.
Will this optimistic trend continue or start to atrophy? We will be able to verify this in 2025.
For now, the top application areas for generative AI identified by Deloitte also reflect McKinsey’s previous AI research, putting cybersecurity at the top of the application investment chart. This was good to see, given concerns about data poisoning risks in AI and deep fakes, we need more innovations like Troj.ai to integrate the brain power of the security manager (see my blog here). Sales, customer service, product development, and supply chain application areas are in close succession with promising generative AI applications of value.
The study further found that talent, governance and risk remain a priority: 41% of executives said their organization was little or not at all prepared to address talent concerns related to talent. adoption of generative AI, while 22% considered their organization highly or not at all prepared. very highly prepared. Similarly, 41% of executives indicated that their organization was little or not at all prepared to address governance and risk concerns related to the adoption of generative AI, while 25% considered their organization as highly or very well prepared.
What is imperative is to ensure that organizations adopt AI to learn, experiment and develop maturity.
I’ve been designing and building complex AI models and AI software products across various industries for over 15 years now, and what I’ve learned most is that AI needs constant maintained. It’s often like a little child wanting to stand on their own two feet, but the value depends on the data and the parenting around it. It’s not something that leaders should give up easily, but… the sad reality is that between 60 and 75 percent of AI models are never maintained.
So where is the ROI for AI? When will we reach Gartner’s trough of disillusionment?
In any major transformation opportunity, the responsibility always lies with executive sponsorship, leadership and serious 360° capability building. If companies are unable to recruit the right talent, it is essential to ensure that they collaborate with proven talent. Some companies like Purolator have recently outsourced their AI analytics to Deloitte and Canada Post recently moved the entire Innovapost (IT function) to Deloitte. I expect AI outsourcing to continue to accelerate in difficult build, buy, or partner decisions.
Before concluding this article, what I found disappointing about the Deloitte report was not strongly highlighting the crisis we still face in data integrity and the incredible costs we have in data management and waste in data traceability practices that we see across all of our clients. practices.
Before we can really move forward and realize the true promise of generative AI, we need to make sure our databases are in place.
Be wise and make sure to inspect the health of your data before you go on the gold rush only to find that your pan of gold is empty.
Sources: