Artificial Intelligence as a Service (AIaaS) on cloud promising next industrial revolution
Even though cloud computing has existed since 2000, It only emerged as one of the leading technologies in 2008 by adapting to existing technologies such as virtualization, service oriented architectures.Today we see most companies migrating to the cloud by moving their services to Amazon, Google, IBM, Microsoft and a variety of other cloud providers.
Artificial Intelligence as a Service (AIaaS):
Kevin Kelly a digital visionary on TED says “So we have many ongoing tendencies right now, and I think one of the chiefs among them is this tendency to make things smarter and smarter. I call it cognifying — cognification — also known as artificial intelligence, or AI. And I think that’s going to be one of the most influential developments and trends and directions and drives in our society in the next 20 years” He concludes saying “The most popular AI product in 20 years from now, that everybody uses, has not been invented yet. That means that you’re not late”
Before going over facts on how AIaaS on cloud promising next industrial revolution, I would like to shed some light on basics of AIaaS on Cloud.
(Figure 2: Theory of AIaaS on Cloud Source [2])
Cloud AI can be categorized into 4 quadrants in the 2×2 matrix. On Y axis we have the type of client involved which can be physical or software application. On another axis, we have the type of data upload which can be structured such web transactions, stock levels, GPS coordinates and other sensor reading. Human beings are capable of processing structured data but tend to become overwhelmed as data becomes large, at this level AI can process functions with relative ease. Unstructured data include audio, digital photo and video all of which in earlier was difficult for computers to make sense of, but today it is possible with cloud AI services can able to make sense of unstructured data and respond.
With this, I would like to go over current updates on companies adopting the cloud AI services in their business and how this has improved their overall productivity
Highlights of Google Cloud Next ’17 – Day 1 Keynote
Diane Greene, SVP of Google Cloud started the keynote with “Cloud is where a lot of digital revelation is going on in IT industries.” Some of the key points from her speech are
- Google Cloud Platform (GCP) inbuilt with AI and ML engines is capable of finding previously hidden patterns in data to predict newer results which are helping in business productivity.
- Datacenters are now equipped with new generation TPUs which is now capable of handling deep learning problems.
- GCP is now having world class infrastructure with AIaaS features
- Michael White CTO, SVP Disney Consumer Product, and Interactive media mentioned that they have moved software development to QA on GCP, and now all their top ranking games run on GCP. With this, as a software development point of view it frees up their developers and resources, which helps their developers to focus on more innovative work and able to take ML mindset on their road to AI, they now strongly believe this will enhance their overall business
- Bernd Leukert, Member of Executive Board SAP says, “SAP has more than 345 thousand enterprise customers, 76 % of the world transacts revenue uses a SAP system. Now “SAP HANA” which is analytical processing tool used by their 14000 customers is now available on GCP, with this SAP HANA has now had features like advanced analytical processing, spatial crafts, predictive capabilities and text analysis which is mainly because of AIaaS inbuilt in GCP”.
- Home Depot yet another firm moved their IT infrastructure on GCP; they are now expecting the retail market to change in next three to five years more than it has since the home depot was founded nearly 40 years ago. They have around 2000 stores in North America which are now running e-commerce infrastructure by making use of AIaas on GCP
- HSBC is one of the large banks in the world established 150 years ago has moved its 93 PB of data onto GCP and running their data analytics tools on GCP. Darryl West group CIO mentioned GCP equipped with AIaaS helping them in important and critical problem solving such
- Anti-Money Laundering (AML): AIaaS APIs are helping to identify AML activity and reduce false positives
- Finance Liquidity Reporting: with AI and ML support on cloud liquidity processing has reduced from 6 hours to 6 mi
- Some other services they are running on cloud are Risk Reporting, Risk Analytics, and Valuation services
- RT Pittman, chief product officer from eBay says “Outsourcing your IT infrastructure and concentrate on your innovation is a strong growth phenomenon”. He says after they moved billions of live listings, 22 years of old data onto GCP, with AI and ML features they are able to build smarter catalogs, able to list the best item for individual at right place right time
Few more cloud companies which are capable with AIaaS are
Amazon AWS cloud is equipped with popular deep learning frameworks to support over million customers in more than 190 countries. Companies such as Expedia, NETFLIX, courser, Unilever running their web application on AWS.
NVIDIA GPU cloud (NGC) is the GPU accelerated platform having the ability to rapidly build, train and deploy neural network models to address complicated AI challenges. Nimbix Cloud Platform is featuring NVIDIA DGX-1TM is a cloud built for applications mainly for deep learning, analytics and HPC applications
Microsoft cloud Azure server equipped with Altera FPGAs is now capable of providing some of the AIaaS
IBM Bluemix cloud platform is yet another cloud with AIaaS with features like blockchain for business, augmenting human intelligence, quantum computing platform, and neural networks etc.
Based on the current updates, it is pretty evident that every cloud provider is trying to bring AI into their cloud, because of the advancement in hardware infrastructure; it can be next generation TPUs, NVIDIA GPU, AWS, Microsoft Azure or IBM Bluemix. Each of this is capable of processing some of the complex AI challenges, and hence every industry are integrating high-performance cloud infrastructure and cutting edge services in their IT environment, this is dramatically boosting the retail market.
So I personally feel bringing AIaaS on cloud is definitely a promising step to bring humans and machines to work together which enables new solutions for the existing problems and also opens up new innovative ideas to build better world
[1] http://www.oivindhovland.com
[2] https://www.youtube.com/watch?v=iv7_Bf7TGOQ
[3]https://www.reddit.com/r/timferriss/comments/3giaz6/complete_summary_of_kevin_kellys_ama_on_the/
[4] https://youtu.be/j_K1YoMHpbk?list=PLIivdWyY5sqI8RuUibiH8sMb1ExIw0lAR
10 comments on “Artificial Intelligence as a Service (AIaaS) on cloud promising next industrial revolution”
Comments are closed.
Very interesting post on AIaaS! The examples you gave of promising use cases for AI using cloud services show that access to this latest tech without the need to set up infrastructure can solve many issues. Home Depot using AI is very interesting, because commerce can benefit greatly from predictive intelligence. You mentioned a few current public cloud providers already offering cloud access with AI capabilities including Google Cloud Platform with TPU access and Microsoft Azure with FPGA access. This relates to what was said in lecture multiple times where the latest tech is easily deployed and accessible via public cloud, such as Amazon’s marketplace for applications, and what Jeff Welser mentioned about allowing access to quantum compute via the cloud. Setting up infrastructure for quantum compute currently would not be feasible for many, but IBM allows customers to access it; likewise, many public cloud users would not want to invest in TPUs or FPGAs for their servers, but now they can utilize that infrastructure for AI/machine learning applications. I expect to see a lot of AI advancements now that public cloud providers give access to tools to make development easier!
Users who have LIKED this comment:
Yes Ajit I agree with you, as Kevin Kelly mentioned the most popular AI product which everyone will be using from 20 years from now is not yet invented, by bringing AI on cloud will give opportunity for everyone to take advantage of AI in their product development which brings new innovative solutions for the existing problems and also helps in emerging of most useful AI products that solve many real world problems to make life better and more efficient
Interesting Blog post.
As IaaS, iPaaS, and SaaS offerings become less and less expensive due to distributed computing and lower total cost of ownership (TCO) in the Cloud Computing market, I am curious what the business plan might look like to resell Cloud or Hybrid Cloud services.
– Do you think that companies like Amazon, Google and Microsoft will continue to make use of channel partners to resell GCP, Amazon AWS and Microsoft Azure services?
OR
– Will the Google/Amazon/Microsoft own the market in Cloud and Cognitive Computing?
When it comes to the evolution of AI and ML using Cloud Computing, some analysts that focus on AI and cognitive learning are strong proponents of Google’s Cloud Platform and Amazon AWS. In discussions with other Cloud experts, there seems to be a perception that they’re the only real players in the market.
While some individual and organizations are advocates for Microsoft, there seems to concern around who has access to the most information to grow their AI/ML platforms.
E.g.
Google has access to a significant amount of information obtained from Search, Android and numerous acquisitions like Waze, Nest Labs, bebop, Apigee, and YouTube (https://en.wikipedia.org/wiki/List_of_mergers_and_acquisitions_by_Alphabet)
Amazon AWS has access to a significant amount of information obtained from their online retail platform, Amazon Echo and Echo Dot using Alexa AI, data from connected IoT devices. They have also acquired more than 15 companies in the last 2 years. (https://en.wikipedia.org/wiki/List_of_mergers_and_acquisitions_by_Amazon.com)
Microsoft has access to a significant amount of information obtained from Windows users and Bing. Most of their Big Data is from existing customer data bases (Windows OS/Office 365), Bing Search and from acquisitions like Skype, Nokia and Linkedin. While they have significant investments in organizations like Facebook, AT&T, Comcast and Telewest Communications, they do not have access to information that will facilitate growth for their Cognitive Computing Platform. Still, Microsoft is a strong player, but I think it’s safe to say that Google and Amazon are the top dogs in the Cognitive (AI/ML) space.
– Do you think that Microsoft, or Facebook can catch up with Google/Amazon AWS?
I would be interested to know what strategies other organizations are using to scale their AI platforms and applications to compete with Google and Amazon.
Users who have LIKED this comment:
Hi Christian
For your first question on whether cloud providers channel partners to resell their services or they own by themselves.
Here I see AIaaS is utility someone can make use and rebuild their own product, maybe we can compare AI to electricity it will transform everything, even though electric companies did not make more money but the gadget makers did.
If someone wants to build their own AI product, apart from their manpower thinking towards ML to drive AI on road, they need infrastructure and also they need to meet commercial grade AI standards. So I personally feel not all the companies can invest their time and money on building commercial grade AI infrastructure, and definitely, they will tie up with existing cloud providers to avail the AIaaS into their existing products and also for new innovations. Also Cloud providers along with providing AI as a service they will also try to come up with their own product solution which definitely brings them more revenue
And on your last question, on will Microsoft or Facebook can catch up Google and AWS. Here we see each one of them approaching different technology to build AI infrastructure, for example, Google TPUs are ASIC (application specific integrated circuits) they run at much higher clock rates compared to Microsoft Azure which is Altera FPGA based (Field Programmable gate array). But FPGA can be reprogrammable, that means every time you have new neural network approach to solve existing problem or bring new solution you can just reprogram the FPGA, so time to market is much faster compared to ASIC, because ASIC chips need to replace with new one and it may take at least 6 to 9 months for the development cycle to fabricate new chip . Having said that we can not say which one is best only time should decide in near future. with this I can say every company including Google, Amazon, Facebook, Microsoft, IBM are still at the beginning stage of bringing AI features on their infrastructure. As Kevin Kelly mentioned “The most popular AI product in 20 years from now, that everybody uses, has not been invented yet. That means that you’re not late”. So I personally feel we have long way to go and it is too early to judge who is going to emerge as leader in this platform
A well written post.
I agree with the technology being very promising. Moreover it is already in use in our day-to-day life. We consume a lot of AIaaS and don’t realise.
Although I can’t get technical on this topic, because you need to know a lot about cloud, it’s services and architecture and I don’t so., On a general note I can say that how ERP was revolutionary, followed by AI, Cloud I believe AIaaS will become an integral part of most companies & this process has already started.
Since most of the companies are already on cloud and they have outsourced or hired people to get the automation part of it done, Instead if they use AIaaS will reduce a lot of cost for the companies making them more efficient.
On the contrary, there can be disadvantages of cloud, if the server goes down. (Based on my previous comment)
Hence, I feel a lot more research could be done on the pros and cons of each company that provides these services so that the experience can be optimized.
Users who have LIKED this comment:
Hi Pratik,
Yes, you are right, every technology comes with pros and cons, and AIaaS on cloud also comes with same. I totally agree with your point, it is always necessary for any new firm trying to bring AIaaS services on a cloud into their business must do prior research on disadvantages involved with cloud and they should have solutions to resolve it. Below article talks some of the interesting concerns on adopting cloud
https://www.accenture.com/_acnmedia/Accenture/Conversion-Assets/DotCom/Documents/Global/PDF/Dualpub_25/Accenture-cabu-position-paper-cloud-concerns-us-web%20(1).pdf
Users who have LIKED this comment:
Great discussion !!! i think AIaaS is the most promising service of the most SaaS service provided till date and startup would benefit the most from this as AI tech which is based on large amount of data is not very simple to obtain and run due to lack of commercial grade infra for smaller companies… with tech changing faces at lighting pace it is difficult for the newbees to keep their infra updated while disrupting the market with their innovation and AIaaS would be the most comprehensive platform to suit their needs considering the cost, resources and time to be invested… but all said and done new development and data privacy still remains a big concern specially when this market is dominated by a few player giving them virtual monopoly in the space and it would be very difficult to prove manipulation given the immense amount of complication and limited knowledge of the judiciary. To debate Christian’s question on channels – i think channels such as System Integrators would continue to have a good opportunity to offer AIaaS platform services to companies as they are already into reselling other products offered by AWS and GCP and the market is huge for Amazon or Google or Microsoft to directly tap into when it comes to the last mile, also there would be another opportunity for technology consulting companies to use these platforms for their client needs as most of the startups dont understand or have limited resources/knowledge to integrate or use platforms such as GCP or AWS.
Users who have LIKED this comment:
Hi Pratik
I agree that security concerns preventing more business adopting the cloud technology, based on current advancements in industry, I personally feel there should not be that difference in security in cloud v/s in premise, as we see its a win and win situation for cloud service provider and user to keep the data more secured in order to build next generation products and sustain in their respective markets. Google has more than 700 employees in their security team, which is larger than any mid sized start up. So even though security is a concern which definitely it will not become a big barrier for the world to adopt cloud in coming days
Users who have LIKED this comment:
I agree with your viewpoint that AIaaS is going to be one of the most influential products in the coming decades. Cybersecurity will be critical for companies to adapt to public cloud; and AI in cybersecurity would help companies efficiently monitor risks and their intellectual properties. As your article presents that AI has advantage in processing unstructured data, a recent acquisition of Amazon reflects the increasing trend of AIaaS.
Harvest.ai provides service to monitor virtual data in real-time fashion and identifies risk to the business of data shared outside the organization.
https://techcrunch.com/2017/01/09/amazon-aws-harvest-ai/
Another promising cybersecurity-AI companies also demonstrates the rising importance of AIaaS. SkyHigh, provides Cloud access Security Broker offerings to help companies safely adopt cloud services without the need for developers to write a single line of code. It has raised total funding of $106.5million.
There is also saying that cloud would be how most companies engage in AI; so AIaaS is definitely an interesting field to learn more about.
Yanxiu Chen, student in MS&E238A.
Users who have LIKED this comment:
Thank you Ramdev for your insightful blogpost on an additional XaaS.
I think AI technologies are powerful tools to enhance our reality and the way we can detect, understand and solve problems.
As I am a little bit familiar with the fourth industrial revolution, I was wondering due to your headline, if you think that AIaaS will support the industrial revolution nowadays or if you are talking about another one in the future?
I believe AI is a good way to improve the digitalization of factories and to get a hold onto the tremendous amount of data, which needs to be analyzed, to receive insight into yet unknown, underlying pattern. Due to the combination of AI and cloud distributed data sources can be aggregated and handled together.
Users who have LIKED this comment: