Salesforce AI & Einstein: Technology You Can Trust?

“Technological progress is like an axe in the hands of a pathological criminal.” -Albert Einstein

An interesting take, but say it ain’t so! Previously we have written on top considerations to take into account before implementing AI and Salesforce. In this article, we give you the rundown on the Einstein 1 Platform and how it has the potential to bring a sense of trust into the realm of AI.?

You would have to have had a very sleepy 2023/ 2024 to have missed the hype around Salesforce AI integration. Artificial intelligence in Salesforce is the hot topic and one we are keen to explore, question, and more fully understand. What was known in July 2023 as AI Cloud has now been rebranded as “Einstein” - a riff on everyone’s favorite 19th & 20th-century physicist. And while Salesforce hasn’t pioneered any theories of relativity just yet, they have hopped on the AI bandwagon faster than you can say “space-time continuum”. Given the pace at which these innovations are being rolled out, and the fact that we are being carried along with them, it’s useful to sometimes pause and figure out exactly which rocket we are on. And what planet we are headed to!?

Hold on tight to find out more…

Salesforce AI Tools: The Trust Layer?

When scanning AI news, articles, and opinion pieces, most doomsday prophets are worried about two major things. The first is that AI will backfire and spread a bunch of fake news causing wars, societal polarization, and the like. The second concern is that data security will be compromised… BIG TIME! In a worst-case scenario, once a bot stores your customer information, could it regurgitate it anytime and anywhere??

While Salesforce wasn’t the first to the finish line in developing AI tools and Salesforce AI products are not necessarily the most cutting-edge on the planet, the CRM giant has perhaps found its niche in bringing trust and a sense of ethical safety to the drawing board. This is because AI grounded in the number 1 source of customer data is unlikely to become a source of untrue information. Furthermore, artificial intelligence in Salesforce is built on the Einstein Trust Layer, which puts responsible practices front and center. While the features and architecture are still being fully worked out, it should provide a solid foundation for future AI developments.

Dynamic Grounding

If you haven’t yet taken any Udemy or LinkedIn learning courses on how to craft good AI prompts, then don’t sweat it. All you need to know is that prompts work better with LLMs when there is more context provided. Einstein Trust Layer provides just that through multiple and dynamic grounding so that AI has context on your organization.

Zero Data Retention and Data Masking

And while Dynamic Grounding helps LLMs figure out what you want, safeguards like Zero Data Retention and Data Masking keep your data secure. Data masking actually empowers you to flag sensitive data (PII-Personal Identifiable and PCI-Payment Card Industry Information) to AI so that this can be masked accordingly.?

Zero Data Retention means that all the grounding and contextual information AI prompts receive is not stored, nor is their AI output.

Toxicity Detection

Toxicity means that AI doesn’t get to decide what’s right or wrong. Employees can mark prompts and AI content as “toxic” if it has the potential to harm or expose data inappropriately. Word on the street is that this feature is still in the process of being released but it can’t come soon enough! This is yet another example of an innovation racing to keep pace with Salesforce’s AI roll-outs.

AI Acceptable Use Policy

Something that Salesforce has developed is an AI Acceptable Use Policy. This is probably something every organization will have in the near future and Salesforce has done a pretty good job at blazing a trail. For one, the PDF is less than 3 pages, so there is a chance that employees and users will actually read it! It makes sure to stipulate that its AI is not to be used for violent purposes, political campaigns, exploitation, making automated legal decisions, or deceiving the general public. This document is a quick and easy read, so we suggest you check it out here . The ethical guidelines in this Acceptable Use Policy are well worth upholding and well laid out. What remains to be seen in the next 5-10 years is if legislation of any kind has the power to regulate the unstoppable force that is AI…

Salesforce AI Research

The values behind AI research within its organization are one thing Salesforce can control. Here, they are committed to stated values of responsibility, inclusivity, accountability, transparency, and empowerment. In this sense, the fundamentals of this research are true to the spirit of Ohana and the sense of community that has bound so many of us to the ecosystem in the first place. And if all organizations were to put these values at the forefront of their AI research, this world (and the next one) wouldn’t be such a bad place.

Potential Pitfalls of Salesforce Einstein Trust Layer?

If there is an amber flag here, it is probably that the Einstein Trust Layer is still maturing. While features like Toxicity Detection are not fully rolled out, Data Masking availability is also dependent on geographic location, feature, and other factors. Furthermore, it is not as though Salesforce itself is immune to data security issues in a world mired with threats. While Salesforce does meet the highest standards of compliance, it has the foresight to know this isn’t always enough. This is one reason why it has spent more than 18.9 million USD funding ethical hackers who have detected more than 30,600 potential vulnerabilities since 2015.?

Salesforce Einstein Trust Layer: Wrapped

And that’s a wrap! Unlike some other corporate giants, Salesforce is committed to the safe and responsible use of AI. This is evidenced in its strong commitment to ethical guidelines and the Einstein Trust Layer. Will this be enough in a rapidly changing world? Watch this space!