Laurent - stock.adobe.com
Content is king, but context is queen
Context is having a major impact on those making the buying decisions as artificial intelligence spreads its reach even further
The one technology market that seems immune to Covid-19 is artificial intelligence (AI). In fact, the virus has only made it stronger.
Everyone in technology expects flexible working to continue after the coronavirus crisis has died down. That’s if the suffocating atmosphere of fear ever does lift. Research by Amsource says 97% of tech professionals believe remote working is here for good and 75% of tech leaders are trying to implement this.
Now that everyone is working and studying from a distance, there has been a substantial increase in automated management of business. There has to be, because there’s far too much to do otherwise. However, some IT operations staff still see AI as a threat, rather than an opportunity to offload the boring, repetitive tasks onto an automated underling.
The latest AI software market forecast from analyst Omdia identifies different growth patterns that this market might undergo. So it’s not a question of if AI will sell, but how furiously. It is obvious that many industries won’t be able to survive without some “assistive intelligence” from machines.
Take network operations people. They are rushed off their feet. Telcos and datacentre operators will need AI to handle the huge workload involved in transforming their infrastructure. All those intransigent silos of data and the rationalising of the fiddly cabling legacy will be impossible to manage without drastically cutting the labour intensity of these jobs.
And that’s before they start thinking about building out their 5G networks. European telcos alone will spend $2bn on AI in 2025, roughly 10 times what they spent in 2019 ($217m), predicts Aditya Kaul, research director for AI and intelligent automation at Omdia.
Then there’s the health sector, retail and financial services to be going on with. All are currently struggling with managing employees who are working from home and doing things with their office computer that would normally be “not suitable for work” – which exposes their employer to massive danger.
The insecurity of the workforce and the instability of the customer base could be more lethal than Covid for many companies.
Retailers will need AI to help them cope with everything from logistics to marketing. Omdia’s report lists some of the project opportunities as “stock, supply chain and inventory management, image recognition, visual search, content management and advertising”.
Highly complicated
The problem with AI is that it’s extremely complicated and everyone is desperate to get these projects completed yesterday.
However, it’s a frighteningly complex subject, and you pity the poor buyers who have to research it.
The other problem with AI is that there’s a lot of data around – but without context, it’s useless.
“People used to say that content is king. But context is queen,” says Amy Holder, AI programme manager at Neo4J.
The queen, says Holder, is the one who rules the board, while the king plods from one square to the next. The queen of the house usually makes all the buying decisions, too.
Neo4J creates graph technology that is designed to make it easier for network operations staff to see the patterns emerging in data. By visualising links for those who supervise AI (when machines need it), Neo4J makes it easier to create the logical building blocks required.
The channel has a crucial role to play in the AI market by humanising it to make the decision-making clearer and the execution easier, says Holder. In short, AI buyers need some assisted intelligence on the AI market.
Market is maturing
Meanwhile, recent consolidations might be a sign that the AI market is maturing.
Data science platform maker Iguazio is partnering with consultant SFL Scientific to make it easier to install the AI functions it specialises in, such as predictive maintenance, instant recommendations and fraud prevention, using both AI and machine learning.
SFL Scientific has been Nvidia’s Service Partner of the Year for the past two years, and Iguazio is one of the first partners in the Nvidia DGX-Ready software partner programme.
AI won’t do anything valuable with your data unless you start by transforming your workforce, production processes and operating models. Then you can think about modernising your critical applications in their infrastructure and architecture, says Iguazio CEO Asaf Somekh.
That is a tremendous task and far too daunting for most people, so Iguazio simplifies it by automating the machine learning at every stage of the project. It creates a sort of pipeline for the process that starts with data collection, then proceeds through training, development, deployment, monitoring and management of the models in production.
Powerful machinery
In the new partnership, SFL Scientific will offer data strategy and develop algorithms for datacentres and cloud infrastructures, while Iguazio provides the powerful machinery to refine the processes. It’s all about making AI more accessible, says Somekh. Designing for ease of use and making the technology more widely accessible is anther sign of market maturity.
“Until now, AI solutions at scale, harnessing data from millions of end points or running multiple models simultaneously, were a luxury only afforded by tech giants,” says Somekh.
Iguazio’s data science platform is cloud-based and built over Kubernetes, so its capacity can be ramped up at will. It uses open source tools popular with data scientists, so they enjoy the freedom of working unconstrained, and this openness means they can concentrate on better resource allocation, with options such as GPU sharing.
“Most AI models never make it to production,” says Somekh. Iguazio aims to create a process with fewer points of failure, he adds.
This is where machine learning operations come in. They can break the deadlock in the data science lifecycle and join the silos between data scientists, engineers and DevOps teams. Having the right technologies in place is easy – the tricky part is building the right processes so that data science teams work in harmony with the engineers and the DevOps teams, which is rarely easy.
You can try to build these complex systems yourself, but if it’s not at the core of your company, you’re embarking on a pointlessly painful journey of discovery. Why build your own car for very trip when there’s a cheap taxi service?
Many get lost on the journey between project and production, says Somekh, adding: “That’s where we can give teams a return on investment in a fraction of the time.”
