When cloud computing is mentioned, the names of the behemoth providers like Amazon’s AWS, Microsoft Azure, and Google Cloud come to mind. These providers are now beginning to facilitate the increased adoption of AI capabilities by opening their platforms through cloud computing AIaaS – ‘Artificial Intelligence as a Service’.
As of 2021, it’s fair to say that the explosion of new technology products and services has been largely fueled by the ubiquity of cloud computing. We’re still in the early stages of what this has to offer; yet this same ubiquity is fueling the rapid and continual rise of artificial intelligence.
Cloud computing’s role in AI adoption
Cloud providers initially facilitated businesses in their implementation of AI by offering big data storage solutions and cheap computing power. These offerings allowed companies like Netflix the ability to create smart products and services on the AWS cloud.
The capabilities of the cloud have also birthed a collection of other industry-disrupting startups like Grammarly, a web based text content editing tool that uses AI-powered machine learning and deep learning technology to aid the writing of its users.
Cloud service providers initially offered enormous amounts of computation power and storage capabilities that allowed businesses to explore and evolve AI capabilities. Yes, that was – and is still – a great service offering, but providers have now gone a step further and packaged AI models on top of their cloud offering in a way that will potentially transform business at every level.
Over the years cloud providers introduced SAAS, PAAS, and IAAS. Businesses now have access to AIaaS too.
AIaaS is essentially about making AI ubiquitous in the same way that cloud computing technology is ubiquitous. AIaaS seeks to offer AI services to a customer base ranging from government agencies to small startups.
Whether it’s Joe Blogs with an idea he wants to develop in his bedroom or a large multinational corporation wanting to process data to improve efficiency, the democratisation of AI technology through cloud computing will make it easy for businesses to trial projects requiring underlying AI components.
Cloud computing Aiaas and AI
Let’s be honest, the average person doesn’t know what cloud computing really means, nor do they understand how it works. Tech consumers are satisfied knowing that they can back up phones, save images or files to Google Drive or Dropbox, or access a banking app without having any inclination that it runs on the backbone and ubiquity of cloud computing.
This level of abstraction potentially increases when AI is thrown into the mix.
Common hardware devices like desktops, laptops and mobile devices do not have sufficient power to proficiently run the vast majority of AI applications. Although the advantage of running an AI model on a local device like a laptop or desktop is not having to connect to the internet, cloud providers offer GPU instances that deliver greater amounts of computing power.
This computation power inevitably enables AI algorithms to process data faster and with more accuracy. Nevertheless, it is possible to run a hybrid model where less computer-intensive tasks are allocated to a local device.
It’s safe to say that without cloud computing, artificial intelligence would not be where it is today. Cloud computing certainly aided and fast-tracked the tenacious work of Professor Geoffrey Hinton’s research on deep neural networks. Known as one of the godfathers of AI , Geoffrey Hinton has been instrumental in transforming AI theory into a working practical reality.
Swami Sivasubramanian speaks about AI algorithms, like deep learning neural nets, and explains that they were devised and created as far back as two decades ago. But the catalyst for increasing adoption and application in recent years is the easy accessibility of specialized computer infrastructures, such as GPUs, specialized CPUs, FPGAs (field programmable gate arrays) and others.
The Cloud Computing ecosystem is favourable for building SaaS business models that run off AI. Furthermore, the magnitude at which data sets are being generated and stored, combined with computation power, makes it possible for a business to conceive and create ideas that were impossible only a few years ago.
The expensive conventional approach to AI development
Harvard Business Review predicts that over the coming decade, AI will contribute over $12 trillion dollars to the global economy. This presents a great opportunity for business. However, companies are presently failing to fully harness the promising potential that AI offers.
Pfizer, a US-based multinational pharmaceutical company had to end their attempt to use IBM Watson to rapidly identify new drugs. The project expenditure ran into the multi-millions but had not delivered anything that could be classed as a breakthrough or revolutionary.
The Anderson Cancer Center in Texas also had to decommission its AI project because of the inability to create a moonshot breakthrough. That being said, both projects were able to achieve small breakthroughs relating to the optimisation of tasks that ultimately increased efficiency.
Pfizer, for example, had success with about 150 smaller projects, including one that could identify patients that were likely to comply with a prescribed medication dosage over the course of a predefined time.
Skills required to harness the potential of AI
Developing algorithms that turn AI ideas into viable business outcomes and solutions requires experienced data scientists and data engineers. They either need to possess specific domain knowledge or partner with domain knowledge experts.
This combination of skill sets can be used to create and then iterate until the desired outcome is achieved. However, it is not always financially feasible for most agile businesses or entrepreneurs, and this process becomes even less viable when you factor in the amount of time required to get a business up to speed to deliver on a value proposition.
Organisations that want to explore implementing an innovative AI end product or service will need to factor in the financial obligations required to launch and maintain a project of this nature. If this scenario depicts how AI is to be approached, then AI exploration may currently be an unfeasible proposition for most businesses.
How cloud computing offers AI services to make AI more accessible through AIaaS
I’m an iPhone user and sometimes find myself trying to push the limit of Siri’s (the voice-activated apple AI) capabilities. I discovered the marvel of Siri accidentally and often try to find new ways to use it. I was impressed when one day, I was unable to turn on the TV whilst cooking dinner because my hands were messy.
I had previously synced my TV to my phone, so I thought, what the heck, just give Siri a try. “Hey, Siri, turn on the TV.” To my amazement, my Apple TV, which is synced to my iPhone, turned on my smart TV. This backstory is relevant because AIaaS has started to open doors for businesses to harness similar technology that currently runs under the hood of products and services like Google’s Home, Apple’s Siri and Amazon Alexa.
Cloud providers have made it easy for businesses to adopt AI by allowing enterprises and startups to leverage their investments into the research and development of AI and its subsets like deep learning, reinforcement learning and intelligent automation.
One of the major prerequisites required to create useful AI is a huge amount of data. This data needs to come from somewhere, and the internet solves that problem. It also needs to be stored somewhere, and cloud storage solves that problem. Cloud computing provides computation power to run AI algorithms, making the fusion of AI and cloud computing seems like a perfect match.
Digitization of data
The digitisation of literally everything means that the magnitude of data being collected is simply incomprehensible for humans to sort through and decipher. The thought of attempting this feat can become overwhelming.
In addition, the amount of data being generated is constantly increasing and making the cloud its home. Furthermore, most businesses don’t have the tools to interpret powerful insights or recognise the opportunities that the data may contain.
This is where the benefits of leveraging AIaaS come into play. AI and machine learning rely on data (and lots of it) to effectively complete an assigned task or to effectively create many of the life-transforming products and services we are now seeing across applications.
This growing trend of AI-backed applications or PaaS with artificial intelligence components has created new business opportunities for cloud providers. The usual suspects are contesting the AI services land grab: Amazon, Google, Microsoft, Apple, IBM, and Salesforce. You then have this pack being tailed by numerous other startups.
The emerging prevailing thought of how the application of AI and its fusion with cloud technology will play out goes something like this: Cloud providers will serve as the storehouse for AI capability and functionality, which enterprises and entrepreneurs will be able to access whenever they have a need.
AI everywhere because of cloud computing AIaaS
Microsoft has plans to “democratize AI” similarly to its involvement in cloud computing. In short, this means that AI development capabilities will become available to any person or company. At the same time, Amazon is following this trend and has Amazon Lex, which is the engine that powers Alexa. Amazon also has Amazon Rekognition, which has been adopted by US law enforcement bodies.
The successful fusion of cloud computing and AI
A by-product of this fusion is the increasing number of companies and startups that have taken the bold step to build their business models around the fundamentals of AI technology. Below is a list of notable AI startups:
- Tempus
- Ascent
- DataRobot
- Freenome
- Grammarly
- CloudMinds
- H20.ai
- Nauto
- OpenAI
- Sift Science
- SoundHound
- Vicarious
- Zoox
- Zymergen
The trend saw AI startups achieve a record $2.4 billion of funding in Q2 of 2019 and it is projected that the market will be valued at $118.6 billion by 2025.
Cloud storage and computing capabilities combined with AI give businesses the ability to capture maximum value from the data they collect. By feeding this data into AI models, insights gained can help businesses become more efficient and work smarter or even create new business models.
It’s easy to understand why businesses want to jump on the AI services bandwagon. AI and ML solutions promise speed, efficiency and optimisation. This means that tasks that would have usually taken a dedicated team several weeks to accomplish can now be completed in a fraction of the time.
An example of this can be found at the ad tech company GumGum, which uses AI to optimise the creation of advertising. Prior to GumGum’s introduction of AI image recognition models, the business used to have brand images edited by hand, frame for frame by humans.
GumGum’s CEO, Ophir Tanz went on to mention in his interview for PC magazine that this is no longer the case, as what used to take a team of 10 people several weeks to do can now be done in a matter of seconds.
The adoption of Cloud AIaaS and AI services
Cloud computing has also revolutionized the way business is conducted in several ways.
- The cloud has dramatically reduced the amount of time required for a business to take an idea from initiation to lunch.
- Small startups have become nimble and can quickly prototype business ideas.
- Small business entities are able to compete in the same arena as larger corporations.
- Global corporations can become agile to the extent that costs are saved across departments.
Cloud providers invest heavily in research and development and are obviously motivated by long-term profitability. Nevertheless, their contribution to the AI ecosystem encourages innovation. And over the course of the next few years, we will see transformation across industries and sectors because products and services will be built on top of AIaaS.
Cloud computing offerings like IaaS, (infrastructure as a service), PaaS (platform as a service) and SaaS (Software as a service) have all contributed to how businesses operate. It’s in this same vein that cloud providers seek to begin offering AIaaS (artificial intelligence as a service).
Cloud computing and AI have been fused together to allow access to AI on a consumption request basis. This means that AI resources can now be used on a per-need basis rather than companies and entrepreneurs having to outlay the operational expenses for technical resources required to get ideas off the ground.
By offering AI services via AIaaS, cloud business plan providers are comparable to torchbearers, signalling the availability of AI services and inviting willing participants to create innovative products and services.
Cloud providers have also made experimenting with AI a reality for businesses and entrepreneurs looking to leverage AI for robotics, automation, sales, analysis and process, customer service, medical diagnosis and predictive analytics.
How AI can be applied through AIaaS
AIaaS opens up countless opportunities, ranging from the optimisation of mundane manual processes to the acceleration of research in science and medicine. Let’s not ignore facial and voice recognition, fraud detection and a plethora of other possibilities. The only limit to what can be achieved through AIaaS is the creator’s imagination.
Domain experts across industry sectors are familiar with claims of how AI has the potential to transform their industry. It should be said that AI is not a magic wand. Rather, it’s the experience and knowledge of domain expertise combined with the fusion of AI or AIaaS solutions that will create the sweet spot where innovation happens.
In an interview with ‘Great Learning’, Bindu Reddy from Abacuas.ai presented some key points to consider when using AIaaS options.
- Where are the pain points within your industry, business or company, and where are you collecting and storing data?
- Secondly, what can that data tell you? What insights are the data giving, and what patterns are formed from the data?
- Last but not least, what predictions can be made from the data that has been collated?
The supply chain is a good example to explore. It is based on companies having to predict how much of something they want to make in the future, which can eventually be sold at a profit.
The supply chain use case also presents ideal scenarios that can allow AI-inspired optimization to reduce inefficiencies and fulfil demand correctly.
Demand forecasting is a good problem for AIaaS to focus on. If a business can predict how much of a particular product they sell at a particular location or at a particular time, imagine how much more efficient and optimised the company will be. In construction, for example, companies can attempt to predict the optimal price at that materials could be bought and sold.
Businesses and entrepreneurs contemplating AIaaS should avoid the buzz and hype of AI, specifically assuming AI is a magic wand that solves all problems.
The AIaaS value proposition
Although AI will revolutionise business, it is equally important for those looking to adopt cloud computing AI services to be grounded. Businesses need to understand what problem is actually being solved and what value is being generated for the business or being provided to its customers.
There should be a clear definition of what the business case is and how the adoption of AIaaS can uniquely add value to fulfil the business case.
It’s widely agreed that cloud providers are demystifying how AI components can be applied by offering plug and play-like features, which allow third parties to build on top of a provider’s AIaaS.
However, AI should not be considered out of the research stage yet, and businesses will find true value by integrating AIaaS into existing systems or interfaces or in the innovation and application of solving genuine problems.
The USP of AIaaS is that it allows the power of AI to be harnessed without the initial and ongoing financial commitment, which also comes with a steep learning curve.
Fortunately, businesses don’t have to procure the expensive human brain power and expertise of data scientists because cloud providers are now able to provide AI tools on a pay-as-you-use basis.
With the option of AI services, businesses now have the power of machine learning, deep learning, NLP, and other artificial intelligence subsets at their fingertips.
Final thoughts
Cloud computing and AI go hand in hand and complement each other almost perfectly. However, I believe the claim that AI As A Service is a game-changer is totally justified.
Cloud computing and AIaaS will play a significant role in the increased adoption of AI. What business wouldn’t want the ability to quickly analyse data that can influence positive business decisions or warn of issues before they arise?
The convergence of AI technology and cloud computing technology offers opportunities to businesses on several levels.
Perhaps most important is the ability to gain access to predefined and prebuilt AI components. Businesses can cost-effectively access and collate granular data that can be dissected and processed by AI algorithms in real time.
Businesses will also be able to make use of cloud computation power to convert ideas into reality. Along with high-volume data analysis, this has the potential to speed up innovation.
The question then posed to enterprises, startups and entrepreneurs is what ideas they will bring to the table.
