Yes, all professions are interested in generative artificial intelligence (AI). From developers producing code to marketers creating content, many professionals are looking for, and sometimes finding, ways to increase their productivity with this new technology.
And the rush toward AI is impacting other digital technologies. Analyst Gartner believes that more than 80% of companies will use application programming interfaces (APIs), models and generative AI software in production environments by 2026.
Still, despite the hype, many companies are not interested in this technology – at least not officially.
5% of companies use generative AI in production
Generative AI is still in the exploratory stage for most companies, with Gartner reporting that less than 5% of them are using the technology in production.
Lily Haake, head of technology at British recruiter Harvey Nash, assures ZDNET that major AI projects are not on the agenda for the moment with her clients.
“I’m seeing some really impressive small pilot projects, like clients in the legal industry using AI to generate documents to make the organization more productive,” she says. “It’s all very exciting and positive. But it’s small scale. My clients don’t seem to be exploring generative AI to the point of transforming the entire business.”
But 44% of IT professionals already use AI in business
However, it is not because the company has not imposed the use of generative AI that professionals are not already using this technology – with or without the approval of the boss.
A study by O’Reilly suggests that 44% of IT professionals are already using AI in their programming work, and 34% are experimenting with it. Nearly a third (32%) of IT professionals use AI for data analysis, and 38% are experimenting with it.
Mike Loukides, author of the report, says O’Reilly is surprised by the level of adoption.
Companies could fall into an ‘AI winter’ if they ignore the risks
But while he calls the growth of generative AI “explosive,” Loukides believes that companies could fall into an “AI winter” if they ignore the risks and dangers that come with rushing to adopt it. technology.
This sentiment is echoed by Gartner analyst Avivah Litan, who says CIOs can’t stand idly by. “You have to manage risks before they manage you,” she tells ZDNET.
Ms. Litan says Gartner surveyed more than 700 executives about the risks of generative AI and found that CIOs are most concerned about data privacy, followed by hallucinations, then security.
Let’s look at each of these areas in turn.
1. Risks related to confidentiality and data protection
CIOs implementing an enterprise version of generative AI will likely send their data into their vendor’s hosted environments.
Avivah Litan acknowledges that this type of architecture is not new – organizations have been sending data to the cloud and to SaaS software providers for a decade or more.
However, she says, CIOs believe AI involves a different kind of risk, particularly around how vendors store and use information, for example to train their own language models (LLMs).
Ms. Litan says Gartner has conducted a detailed analysis of many IT vendors that offer AI-based services.
“When it comes to data protection, if you use a third-party base model, it’s all about trust, but you can’t verify,” she explains. “So you need to be sure that the providers have good security practices in place and that your data is not going to leak. And we all know that mistakes are made in cloud systems. If your confidential data leaks , the suppliers will not be responsible, it is you who will be.”
2. Risks related to entries and exits
In addition to assessing data protection risks in external processes, organizations should be aware of how employees use data in generative AI applications and models.
According to Avivah Litan, these types of risks cover “unacceptable” use of data, which can compromise the decision-making process, including misuse of confidential data, production of hallucinations and use of intellectual property from another company.
Combine this with ethical questions and concerns that models are biased, and business leaders face a multitude of risks related to data input and output.
According to Avivah Litan, leaders managing the deployment of generative AI must ensure that company employees do not take anything for granted in this field.
“You need to ensure that you are using data and generative AI in a way that is acceptable to your organization; that you are not giving away the keys to your kingdom, and that what comes back is verified to avoid inaccuracies and hallucinations” , she says.
3. New cybersecurity risks
Businesses face a range of cybersecurity risks every day, such as hackers gaining access to company data due to a system vulnerability or an employee error.
However, according to Avivah Litan, AI represents a different threat vector.
“These are new risks,” she explains. “There are rapid injection attacks, vector database attacks, and hackers can gain access to model states and parameters.”
While risks include loss of data and money, attackers could also choose to manipulate models and replace good data with bad ones.
According to Avivah Litan, this new threat vector means that businesses cannot address new risks by simply using old, proven techniques.
“Attackers can poison the model,” she explains. “You will need to build security around the model and that is a different type of security. Endpoint protection will not help you protect the data model.”
What your business needs to do now to move forward with generative AI
This combination of generative AI risks can seem like an insurmountable challenge for CIOs.
However, Avivah Litan says new solutions are evolving as quickly as the risks and opportunities associated with generative AI.
“There’s no need to sit around and panic,” she says. “A new market is evolving. As you can imagine, when there are problems, there are entrepreneurs who want to make money from those problems.”
The good news is that solutions are being developed. And as the tech market stabilizes, Litan says business leaders need to prepare.
“The key advice we give to CIOs is: ‘Get organized, then define your acceptable use policies. Make sure your data is classified and you have access management,” she says.
“Set up a system where users can submit their application requests, you know what data they’re using, the right people approve those requests, and you check the process twice a year to make sure it’s being applied correctly. Generative AI should be used step by step.”
Source: “ZDNet.com”