Readymade AI favored by early adopters

Businesses are using readymade artificial intelligence capabilities as a first step in embracing the new technology, according to Deloitte Insights’ “State of AI in the Enterprise“.

Additionally, the survey shows AI is producing varying levels of ROI among users, with healthcare and fiscal functions yielding the least.

Most everyday users of AI currently access the technology – or rather, its results – via the functionality built into their existing ERP systems, with Salesforce’s Einstein and SAP’s Leonardo namechecked by many respondents.

The executives surveyed by Deloitte Insights were mostly positive about the benefits artificial intelligence could offer, although each of the survey’s findings should be considered as indicative of attitudes among early adopters, rather than of businesses in general. Deloitte stated:

“All survey respondents can be considered early adopters when compared with their counterparts in an average company. All of the respondents’ companies have implemented at least one cognitive prototype or full-scale implementation. In addition, 75 percent of respondents say they have an “excellent” understanding of AI or are experts.”

The survey elucidated the semantic differences among the acronym smorgasbord of AI, machine learning (ML), cognitive learning, deep learning, and so forth, but concentrating on practical takeaways for organizations and decision-makers considering adopting AI – or cognitive technologies – in the workplace.

Companies are additionally developing their own AI functions with partners, using open source and crowdsourced resources, with only eight percent stating they were experiencing a critical shortage of trained (or train-able) workforce.

Google, Amazon and the large cloud providers are currently in competition with one another to capture enterprises hoping to leverage AI.

The cloud giants are keen to offer services which abstract away the AI underpinnings, presenting paths to services which don’t necessarily need a deep understanding of the particular requirements of AI to leverage the resulting output.

This approach is apparently paying off, as, among Deloitte’s early adopters, this form of using AI was ranked second most popular as the routes to the technology.

Education and the public sector were the areas least likely to be investing in AI at present, whereas engineering and production environments were gaining the best results from relatively small investments in time and resources.

While public sectors across the world vary, the one criticism leveled against most (although it is a slur on most large bureaucracies) is that internal processes can be improved.

Of the executives who responded, more were using AI to change internal processes this year than last (36 percent in 2017 and 42 percent in 2018).

Fewer, in turn, were leveraging the emerging technology to develop new products or revolutionize their market (27 percent, down from 32 percent). AI among early adopters, therefore, is settling – at least for now – into a more consolidatory position.

New tech’s impact on the means of production is felt in the industrial internet of things, and AI’s success in this sector may be due to the need to process the enormous amounts of data which such deployments can produce.

Medical and healthcare companies and organizations were reported as being reluctant to deploy AI’s results – perhaps due to the need for legislative and regulatory oversight in that particular field.

Deployments were confined in large part to research and lab environments, although early indicators have shown that prognostic routines are highly accurate, albeit in controlled environments.

Doomsayers’ worries of mass unemployment because of AI-driven robots in the workplace will probably never be assuaged, but the survey’s respondents’ attitudes may bring at least some comfort: only 24 percent were hoping to “reduce headcount through automation” through AI.

Cybersecurity in AI was the highest ranking concern about the new technology, with respondents aware that especially during the learning phases of AI deployment, the potential results of algorithms being shown rogue data among kosher, could be dangerous.