.png?width=300&name=Blog%20Post%20featured%20image%20(8).png)
"Generative AI has evolved to one of the fastest user-adopted technologies, and as regulators and C-suite leaders struggle to keep up, it's causing a sense of discontinuity, confusion, and even a loss of control among employees. As businesses continue to adopt generative AI, leaders must keep employees at the center and help overcome fear-based barriers to usher in a new era of productivity and growth."
— Dan Diasio, Ernst & Young Global Artificial Intelligence Consulting Leader
Let’s step back to that snowglobe, before it got shaken upside down, trace the journey to this point, and immerse ourselves in the current GenAI landscape. Finally, let’s look into the future of a world without limitations, without a snowglobe at all.
Closing the GenAI Adoption Gap: A Call to Empower and Experiment
As we embark on an exploratory journey of GenAI, there’s one thing to bear in mind that we’ll repeat a few times. Companies must use their judgment to adopt GenAI wisely. It's also crucial to start exploring, experimenting, and more importantly — empowering employees to do the same right away.
Source: BCG X
Leaders are adopting GenAI as regular users at an 80% clip — a very high level — while frontline employees are only at 20% regular use and 60% nonusers, falling behind for one reason or another.
More than 50% of executives discourage GenAI use
Source: BCG X
As evidenced in the graph above, the less informed about and experienced with GenAI (GenAI maturity is low), the more organizations are likely to discourage exploring and even using GenAI at a rudimentary level.
The more experience and familiarity with GenAI, the more likely a company is to experiment with it, and deploy it.
This is a true gap that any corporation and leadership team can’t ignore. Regular use boosts optimism and reduces concerns. Corporations must seize this opportunity to enhance knowledge, safeguard against dangers, and not dismiss GenAI.
GenAI optimism grows with familiarity & regular use compared to those who never or rarely tried it
Source: BCG X
To cut through the sediment around the GenAI snowglobe, understanding is key. Adopting, accepting, and adapting to GenAI is not theoretical; it's fundamental both for individuals and company ecosystems.
Groundbreaking by Generation: The GenAI Journey
Generative AI is the next evolution of AI, using sophisticated algorithms based on learning patterns, structure, data and neural networks that allows it to generate content and output.
This seismic shift transforms the traditional linear approach to research, data, insights, and innovation into a dynamic paradigm. In a short span, GenAI has redefined its role as a tool, enhancing our abilities and tasks.
GenAI can generate novel solutions that push the boundaries of design and problem-solving. It goes beyond conventional limits, generating innovative solutions that reshape our perspective on technology's future role.
While the GenAI journey spans decades, our contemporary focus traces its roots back to the 2000s when machine learning laid the foundation.
Source: PWC
The next stage beyond GenAI is Artificial General Intelligence (AGI), which is the AI we’ve been trained to think about from movies and books. The world without limitations, without the borders of a snowglobe, is AGI.
AGI is an AI that can understand, learn and even apply knowledge that will start to resemble and eventually maybe even replicate human intelligence.
Step-by-step to accessibility
The step-by-step advances allowed for AI analysis of data, pattern recognition, insight generation and automation at a scale and speed not known before.
Subsequent advances in AI throughout the last decade, what we now call deep learning, paved the path for concepts such as self-driving cars to speech recognition with new possibilities in the ability to classify, detect, and process.
The foundational models in GenAI are highly-complex systems of machine learning that have been trained on enormous amounts of data — text, audio, images, and combinations of the three — on a scale of millions entering the billions.
The key advancement lies in their adaptability, upgradability, and potential for continuous fine-tuning and tinkering that open up a wide array of capabilities and applications.
Source: McKInsey & Co.
Although GenAI can be used in a wide array of functions, it’s clear that they’re not all created equally. Or rather, the earning potential of them differs drastically. Sales, marketing, software engineering and customer operations are all forecast to drive an impact of $400 billion-plus with strategic corporate use, with product R&D around $300bn — with these sectors representing 75% of the total annual impact projected.
Take those things in tandem, and let’s look at one small use case example of how a practical problem can be solved with GenAI that frees up time, energy and money.
Source: Pilot44
For just about any CPG brand, packaging inspection is a must for quality assurance and customer (and partner) satisfaction. Landing AI leveraged GenAI to implement real-time monitoring at scale, covering the entire production process, with applications and capabilities that are still growing.
It’s a small thing, and yet it shows the potential of how GenAI can create a huge stride forward.
A Large Leap Forward
Now, the spotlight is on GenAI, taking a giant leap with GPT models and their expansive language-based applications. We find ourselves at the intersection of the beginning and middle stages of the GPT & LLM surge in the GenAI era.
Source: SMB Group
From its roots in rules-based systems, GenAI now runs on powerful recurrent neural networks (RNNs), backed by robust hardware and software capabilities. Each incremental addition propels AI forward, placing us both at the beginning and in the midst of the GenAI revolution.
ChatGPT marked a pivotal moment in November 2022. What was once considered cutting-edge in September 2022, is now already common, showcasing the rapid evolution of this technology.
Corporations and GenAI are already intertwined, and those connections will deepen even more as we embark on new breakthroughs shaping the future of innovation.
The Increasing Pace of Progress: ChatGPT Evolution
The first breakthrough in GenAI is considered transformer models in 2017, which led to the 2018 breakthrough of BERT and the advent of GPT 3 in 2020.
Just like the rest of the world, we started playing with ChatGPT and have continued to test out its interface and output capabilities.
As we all progressed in our experimentation and understanding of ChatGPT (and GenAI), it meant becoming more discerning. In other words, it’s all about the output. And, the input.
Scores of suggestions run rampant online, advising us how to better prompt ChatGPT, how to better refine the output we receive and turn that into more methodical input, to again get better output, to again…well, you get the picture.
Sources: Instagram/Alamin + LinkedIn/Mohamed Sakr
Even ChatGPT has evolved, up to 4.0 now as of December 2023, with settings that allow a user to enter in specific information to help refine and direct the output to be catered to who they are, what they want, and more.
Source: ChatGPT
There are layers and layers of knowledge and approaches to consider when trying to fine tune the output you’ll receive, all with a huge grain of salt in considering that there can be inaccuracies, and lots of speech redundancies.
The Gen AI Major Breakthroughs
What do we mean when we say Generative AI? Most people are familiar with the concept of AI, so let’s look at the generative part.
It’s generative because the machine learning algorithm (ML) utilizing neural networks (computer modeling of basic human brain activity and functions) is trained on datasets, or in some cases, thousands, millions, and now billions of datasets, in order to be able to generate new material, new examples, new outputs that were not possible before.
In layman’s terms, it’s generative because it generates new outputs, evolving from AI models of the past that could only mimic and imitate.
With that said, some of these answers were aided by our friendly neighborhood GPT.
GAN (Generative Adversarial Networks)
It all starts here. The driving force behind Generative AI is a type of deep learning model that consists of two neural networks – a generator and a discriminator — that work together in a process known as adversarial training.
The generator creates new data instances, while the discriminator evaluates their authenticity. This iterative process continues until the generator produces data that is indistinguishable from the original dataset, effectively enabling the AI to create realistic and original content.
A GAN could be compared to a supervisor, one that refines its learning without the need for constant human interaction.
The dual networks act as a “split personality” with one viewing the training data in order to generate something new while the other attempts to differentiate between the new output and the input data, according to the BBC.
Source: BBC
LLM (Large Language Models) + NLP (Natural Language Processing)
GPT stands for Generative Pretrained Transformer, and it is the standard-setting large language model created by OpenAI. The enhanced NLP capabilities provide efficient content creation and are constantly evolving with new input of data.
With a little help from the main GPT itself, at the core, an LLM is:
“An LLM agent is an AI system powered by a large language model, capable of tasks beyond text generation, such as conversations and reasoning. Directed through carefully crafted prompts, these agents can exhibit varying degrees of autonomy, making them versatile for applications like chatbots and automated workflows, with their behavior guided by user-provided prompts.”
— ChatGPT
Here’s a digestible snapshot of an LLM in action for the healthcare industry.
Source: Nvidia/Stanford
BERT (Bidirectional Encoder Representations from Transformers)
The key term here is bidirectional, as this system allowed for context understanding in both directions to better comprehend nuances of language, powering sophisticated language translators and sentiment analysis.
StyleGAN (Generative Adversarial Networks for Style-based Generation)
The visual aspect of Gen AI that allows for the widespread generation of realistic images based on wide-ranging prompts.
VAE (Variational Autoencoder)
VAE is a type of autoencoder that adds probabilistic elements to the traditional autoencoder architecture. The model consists of an encoder network that maps input data to a probability distribution in a latent space, and a decoder network that generates new data points from samples in this latent space.
The probabilistic nature of VAE allows it to capture the uncertainty inherent in generating new data.
Reinforcement Learning (RL)
RL is a learning paradigm where agents make decisions in an environment to maximize cumulative rewards. In the CPG sector, RL can optimize supply chain management, pricing strategies, and personalized marketing campaigns based on real-time consumer behavior.
Transfer Learning in Computer Vision
Transfer learning enables models to leverage knowledge gained from one task to perform better on another. In the context of CPG, this breakthrough facilitates image recognition for quality control, inventory management, and shelf optimization in retail environments.
DALL-E
Produces images based on text descriptions, and generally started the rise of digitally-created AI artwork. DALL-E. A name combination of the movie character WALL-E and the artist Salvador Dali, was also created by OpenAI as a variation on GPT-3.
Key Components of Gen AI
There are four components that make up the backbone of GenAI and bring it from theoretical to practical reality. As one sector gets more efficient, it powers the rest of the chain to improved capabilities and ease of use and access.
Source: Pilot44
The specialized chips — namely, GPUs and TPUs — that power ultra-large data processing and crunching that GenAI requires have been very cost and materials intensive, leading to some scarcity and hoarding on the market.
That’s also a big reason why some businesses are using cloud technology to build, refine and execute large AI models because of the scalability and cost-effectiveness of cloud computing computational power. For now, though, the main players are likely to stay at the top because dominance lies with the major cloud players with their comprehensive and crucial access.
As companies become more efficient in developing processors, the progress of GenAI technology will somehow become even more accelerated and widespread.
One area that has potential to start growing quickly are the foundational models. The current foundational models, such as OpenAI, have a headstart in the race, but they’ve also shown the roadmap to replicate toward new innovative models as well.
Finally, with many models hitting the market as closed-sourced, it’s opened the door for model hubs like Hugging Face to create open-source models to support businesses that lack resources to innovate and leverage GenAI on their own.
The Current State of Gen AI
Optimism abounds as corporations and leaders look ahead to 2024.
This is very interesting and shows that the moment for learning and adoption is now because a big majority of global leaders believe 2024 will be the year when adoption grows and turns into growth itself.
With full acknowledgment of the hurdles and barriers, not to mention financial burdens to come, a vast majority of those on the inside are forecasting growth and improved profits in 2024 thanks in large part to the investment in and adoption of GenAI.
In a recent Ernst & Young survey, 66% of 1,200 global CEOs respondents expect some revenue growth while 65% are projecting profitability increases in 2024.
Source: Ernst & Young
In that same study, nearly 2/3 believe GenAI will challenge them to disrupt their own business models in order to maintain and leverage competitive advantages. On top of that, 99% are planning to invest in GenAI.
Source: Ernst & Young
Bridging the Gap: Transitioning from Discussion to Strategic Implementation
It’s being talked about in higher volume than ever, leading to more refinement in exploration across the spectrum of possible GenAI uses and capabilities.
The number of mentions of some form of AI and GenAI is increasing in high-level board meetings, and as we know, board meetings and strategic talk is not cheap.
Source: EY
The corporate relationships with GenAI shows that there is still a large gap between thinking about it, talking about it, and moving forward with robust investment.
Absolutely, it’s worth taking ample time to truly consider and contemplate on making the most-informed decisions possible, but it’s also paramount to begin taking steps and actions toward GenAI being commonplace in what and how you go about business for 2024.
Source: EY
In the chart above, the “in progress” portion has the largest share, but consider what in progress can mean, which could just be talk in meetings about starting to take steps to adopt GenAI, and combine that with the “not started” portion, and on average, only about 40% of companies have completed some level of GenAI action in their organization.
From Generative to GenAI: A Search Journey Reflecting Societal Adoption
GenAI is already a part of our common nomenclature, and the search interest for the topic has stayed at all-time highs for quite some time.
First, the search interest was around “Generative AI,” but as we all became more familiar with it, that shifted to searches for “GenAI” as shown in the Google Trends graphs below.
Search Interest (0-100) for Generative AI: Nov. 1, 2022-Dec. 1, 2023
Search Interest (0-100) for Generative AI: Nov. 1, 2022-Dec. 1, 2023
Sources: Google Trends
The more people have learned about and engaged with Generative AI, the more the term GenAI has become common in our everyday language and society.
Comparing the two Google Trends charts, right around the time the search for Generative AI took a slight dip, there was an increase in search interest for GenAI.
At the beginning of June 2023, Generative AI search interest peaked at 100, and started to drop off slightly right around the middle of the month. At that same time, GenAI search interest started accelerating from about 35 to 50, and has increased ever since, hitting 100 itself at the start of December.
The perceived dip in Generative AI search interest didn't actually signal that searches for the technology, tools and capabilities were dropping — only that people were changing their search to GenAI instead. Taken in combination, this shows that whatever you call it, Generative AI or GenAI, it’s here to stay.
GenAI's Role in Shaping Our Future : Impacting the Marketplace & Workforce
Productivity as a day-to-day thought is being replaced by productivity on larger scales and systems that will fundamentally alter the way we work and the way we live.
The time, energy, money and workforce power that is saved through GenAI advances will be repurposed into new enterprises, professions and quality of life solutions.
Source: MIT
These projections from MIT show that GenAI is estimated to surpass any other breakthrough technology in history. And as we’ve seen, our projections regarding GenAI have drastically changed in just one year’s time, so that 50% could be on the lower side.
Another way to look at this is in terms of the workforce efficiency gains and the economic benefits that will in tandem with that.
Source: GitHub/Pilot44
How will that adoption and adjustment take place? It’s already started, but these percentages are bound to continue fluctuating at least on the margins as we learn and do more in 2024.
Source: Accenture
For instance, more than 25% of technology firms in the US are already using GenAI tools to boost software development, while a staggering 92% of developers are using AI coding tools.
As we covered earlier, the state of predictive models for what will be earned and gained by our immersive adoption of GenAI cannot be entirely quantified, because those prognostications are being updated all the time.
Considering the investment in GenAI worldwide, it’s accelerating as almost all other investments are starting to see a downturn. In the next two years alone it’s predicted that
- 10% of all internet data will be generated by GenAI
- 90% of online content could be generated by AI by 2025
Finally, consider the rate of projection changes since the introduction of GenAI and ChatGPT and how forecasts for technical capabilities have completely accelerated in such a short time, showing that what we thought we knew, what we know, and what we think we know about what’s to come are being altered every moment.
Let’s look at one aspect at the heart of our entire discussion so far, LLM and NLP. Natural-language generation technology fueled by GenAI was once forecast to come into existence between 2040-2060 as recently as 2017.
Now, projections state that the technology is expected to occur before 2030.
Source: McKinsey & Co.
Corporate Adoption & Investment Are on the Rise
There are several hurdles to corporate adoption of GenAI, some of which we’ve covered. While the biggest barrier seems to be lack of knowledge and information, security and risks still factor very prominently.
Source: Statista
Marketing and advertising continue to pace the rate of workplace adoption in 2023, with technology and consulting just behind that based on year-end numbers shown above.
The use of chatbots and virtual agents rose 26% in 2023 while text generation and translation went up 12% as popular use cases, according to Intel.
The number of enterprises using AI overall increased from 48% to 55% from 2022 to 2023, with GenAI investment fueling a large bulk of that increase, with approximately $2.5 billion spent on GenAI by enterprises.
And between July and September, enterprise experimentation and expansion of implementing GenAI increased from 62% to 71%.
Source: Menlo Ventures
Even though adoption is rising, interest and curiosity are still outpacing adoption, which could mean that the lack of understanding we’ve cited throughout is holding companies back from turning that interest into adoption, as evidenced by the figures above.
Source: Ernst & Young
A more recent update on the insights shows that 64% of CEOs are already seeing positive impacts on returns from investing in GenAI, and many more are expecting those impacts to come in the very near future.
New GenAI News By the Day
On Nov. 14, CPG manufacturing platform announced an $18 million seed round of investment, led by Lightspeed Venture Partners.
Keychain wants to help brands find manufacturing partners via GenAI, according to TechCrunch.
“If we can take that process from months down to days, we think we can unlock innovation in product development for this whole industry. That, in turn, will result in healthier, more affordable products on our store shelves”
— Keychain cofounder Oisin Hanrahan
Two days later, on Nov. 16, Siena AI announced that it raised $4.7 million to develop a “empathetic AI customer service agent” with funding from Sierra Ventures, Pari Passu Ventures, Spacestation Investments, Village Global, The Council and OpenSky Ventures, among others.
There isn’t a day that goes by without a new announcement on the GenAI front, from fundraising to implementation, adoption to innovation.
And of course, just recently Google grabbed the world's attention with its anticipated unveiling of Gemini in the first week of December.
Gemini is Google's LLM answer to ChatGPT, and the company claims it already outperforms ChatGPT on 30 of 32 widely-used academic benchmarks used in large language model (LLM) research and development, Forbes reported.
It's the successor to PaLM, and will power Bard & other Google Gen AI tech. It’s highly adaptable, and able to understand text prompts, images, audio & video, and is another demonstration that new capabilities and evolutions are going to increase throughout the GenAI era.
Navigating the GenAI Era: Learn, Adopt, Upskill
All things around GenAI are accelerating at an unprecedented pace. It makes complete sense that some would be hesitant, seeing the power GenAI can have and knowing that adopting too fast, without the proper knowledge, training and protocols in place could lead to negative outcomes.
Yet, the vital aspects of GenAI adoption that any company can leverage are:
- Increased understanding of GenAI, which leads increased comfort and confidence
- Better efficiency and a more intelligently productive company and workforce
- More comprehensive understanding of trends, consumers and competitors
- Access to more data, faster, which will build a loop of even better data coming in, faster and faster
- Cutting monotonous task time by even 80% in some cases
- Stronger ROI on investment now, planting seeds for more robust growth to come
As we visited at the beginning, there’s an opportunity for corporations to shatter the GenAI era snowglobe themselves. The only thing holding that back is knowledge and access.
As a last point of consideration, nearly 90% of people believe they need upskilling to address GenAI in their jobs. Yet, only 14% of frontline employees have had any training or upskilling to address this.
How can a company leverage GenAI if their workforce isn’t familiar with it, much less becoming experts?
Source: BCG X
With a different approach, a well-informed process and methodology to adopt GenAI to advance as a company, newer innovations are possible.
One last consideration, is that high performers in the GenAI efforts have already started looking to the dynamic capabilities to explore new ventures, new business models, and new sources of revenue in the future, as opposed to just trying to save a buck through the technology.
The highly informed already know that long-term gains can happen in a hurry with strong investment in GenAI now. That means investment in the tools and capabilities, in upskilling, in integrations across the enterprise, with a keen eye toward being principled and transparent in GenAI use.
Now that we’ve gone through the looking glass and burst the GenAI era snowglobe, we’ll leave you with one comprehensive picture to consider.
We’re all largely new to GenAI, but we’re incredibly curious and engaged in learning all we can, adopting where it makes sense, and innovating from the inside and outside in order to leverage the accelerated opportunities this dynamic technology presents.
The capabilities are limitless, and we’re excited to keep exploring. If you’re looking for an invested, engaged partner to leverage GenAI in your own company, look no further. We’re ready to shatter that snowglobe with you and see what innovation and ingenuity we can discover together.
