According to a recent survey from SADA Systems of IT professionals at large companies, artificial intelligence (AI) and internet of things (IoT) are the primary areas of focus for enterprise investments in new tech in 2018. Of the 500 IT professionals surveyed, 38% claimed that AI was the primary focus of emerging tech projects, with IoT and blockchain coming in at 31% and 10%, respectively.
IoT connected devices often generate the dizzying amounts of data necessary to train machine learning models. Of companies surveyed, more have IoT workflows already in production than AI. This is because a stable IoT and edge computing foundation are often prerequisites for enterprises to break ground on a machine learning model in the first place, though recent cloud PaaS rollouts of pre-built, adaptable ML models are shaking up the landscape for development.
A study by Vanson Bourne (via Forbes) details that enterprises are primarily investing in AI to improve customer experiences and drive revenue through product innovation. But roadblocks to successful implementation are still aplenty.
Breaking Down Barriers To AI Investment
The question that opens an enterprise’s wallet for AI investment is not just, “How can we implement a machine learning model?” Instead, it’s, “How can we implement a machine learning model that adds value to our bottom line?” Failing to answer this question can lead to wasted dollars and dead-end projects.
Although AI leads the way in new tech investment, there are still plenty of pitfalls for enterprise implementation. In a CIO article, Chris Curran notes that issues with leadership, alignment with business goals and the lack AI-skilled engineers can cause projects to stop dead in their tracks. Companies need to make sure that leadership understands the business case for investment and that a specific department leader spearheads AI development rather than letting small, fractured AI projects flare out across departments.
While AI projects require engineers with machine learning development experience, PaaS providers are doing their best to make AI development more approachable for engineers. Consider data ingestion and the mere challenge of amassing training data. More of the big names in cloud are ramping up support and interoperability for IoT devices, and data ingestion has become easier for the average enterprise running IoT workflows in the public cloud. Couple this with new, pre-built machine learning APIs and libraries, and suddenly, AI projects require less legwork to get off the ground.
On the Horizon: Long-Term Projects Demonstrate Large Investment In Emerging Tech
Investments in new technology require time, testing and iteration. They often navigate murky waters and must adapt week by week to new offerings in the marketplace.
As new, buzzworthy machine learning (ML) libraries and services pop up to support AI and IoT, it’s important for organizations to stick to a clear plan for project goals and deployment. Scope creep in emerging tech development leads to unending development cycles and dreamy applications that never see the production light of day. Compound this with the scarce supply of experts in these emerging tech areas, and it becomes increasingly necessary for companies to retain talent by creating realistic timelines for project milestones. At SADA, we recommend that our clients leverage experts to help them create realistic project timeframes while holding them accountable to the project scope — even if the project scope spans years.
Security And Privacy Are Paramount: Incorporate A ‘Measure Twice, Cut Once’ Mentality
The frantic flurry of news surrounding emerging tech leads companies to believe that they must accelerate development to keep up with the Joneses. In reality, it’s quite opposite. Best practices for security and privacy often lag behind the latest hot-button tech trend. Moving too fast and failing to come up with a watertight plan for security can put hastily built platforms at risk for exploitation. In a rush to be a first mover in emerging tech, it’s easy to overlook essential QA and security pressure testing of new applications.
In a 2018 report from the University of Oxford, researchers suggest that engineers “take the dual-use nature of their work seriously, allowing misuse-related considerations to influence research priorities and norms, and proactively reaching out to relevant actors when harmful applications are foreseeable.” Malintent must be an ever-continuing discussion for teams implementing AI workflows inside the enterprise. One vulnerability in a system can not only spoil the validity of an ML workflow, but it can jeopardize business-critical operations.
This is another reason why organizations might look to hire an expert security vendor as a third-party partner to vet their implementations of new tech.
Will Blockchain See The Same Enterprise Push As AI And IoT?
Although blockchain attracts a large share of attention in tech media, it lags behind IoT and AI in enterprise investment. Blockchain certainly holds promise for widespread enterprise use in the near future, but early adopters must pave the wave for the market to follow.
As more enterprises start to share success stories of practical applications of blockchain technology, and as PaaS providers add better out-of-the-box support, we may see increased enterprise investment by next year just as we’ve seen IoT and AI gain traction this year.
In the meantime, companies should make sure their investments in new tech adhere to project scope. Continuously reevaluate business needs and available supporting technology but ensure that leadership is held accountable to a clearly defined overarching business goal. Doing so will decrease time to production and save significant rework down the road.
Note: Original posting of this was published On Original Source. Link below.
You need to Login in order to vote.Original Source