Head over to our on-demand library to view periods from VB Remodel 2023. Register Right here
As organizations around the globe look to quickly consider, take a look at, and deploy generative AI into their workflows — both on the backend, front-end (customer-facing) or each — many decision-makers stay rightfully involved about a few of the lingering points, amongst them, the issue of AI hallucinations.
However a brand new startup, Gleen AI, has burst on the scene and claims to “resolve hallucination,” in keeping with Ashu Dubey, CEO and co-founder of Gleen, who spoke to VentureBeat solely in a video name interview.
At the moment, Gleen AI broadcasts a $4.9 million funding spherical from Sluggish Ventures, sixth Man Ventures, South Park Commons, Spartan Group, and different enterprise companies and angel traders together with former Fb/Meta Platforms’ VP of product administration Sam Lessin, to proceed constructing out its anti-hallucination information layer software program for enterprises, which is focused initially towards serving to them configure AI fashions to supply buyer help.
The issue with hallucinations
Generative AI instruments similar to the favored, commercially obtainable giant language fashions (LLMs) similar to ChatGPT, Claude 2, LLaMA 2, Bard, and others, are skilled to reply to human-entered prompts and queries by producing information that has been related to the phrases and concepts the human consumer has entered.
However the gen AI fashions don’t all the time get it proper, and in lots of circumstances, produce data that’s inaccurate or not related however that the mannequin’s coaching has related beforehand with one thing the human consumer has mentioned.
One good latest instance, ChatGPT attempting to reply “when has the Earth eclipsed Mars?” and offering a convincing sounding clarification that’s solely inaccurate (the very premise of the query is flawed and inaccurate — the Earth can’t eclipse Mars).
Whereas these inaccurate responses may be at occasions humorous or fascinating, for companies attempting to depend on them to supply correct information for workers or customers, the outcomes may be vastly dangerous — particularly for highly-regulated, life-and-death data in healthcare, medication, heavy business, and others.
What Gleen does to stop hallucinations
“What we do is, after we ship information [from a user] to an LLM, we give info that may create a superb reply,” Dubey mentioned. “If we don’t consider we’ve got sufficient info, we gained’t ship the information to the LLM.”
Particularly, Gleen has created proprietary AI and machine studying (ML) layer unbiased of no matter LLM that their enterprise buyer needs to deploy.
This layer securely sifts via and enterprise’s personal inside information, turning it right into a vector database, and makes use of this information enhance the standard of the the AI mannequin’s solutions.
Gleen’s layer does the next:
- Aggregates structured and unstructured enterprise information from a number of sources like assist documentation, FAQs, product specs, manuals, wikis, boards and previous chat logs.
- Curates and extracts key info, eliminating noise and redundancy. Dubey mentioned this “permits us to glean the sign from the noise.” (Additionally the origin of Gleen’s identify.)
- Constructs a information graph to know relationships between entities. The graph aids in retrieving probably the most related info for a given question.
- Checks the LLM’s response towards the curated info earlier than delivering the output. If proof is missing, the chatbot will say “I don’t know” quite than threat hallucination.
The AI layer acts as a checkpoint, cross-checking the LLM’s response earlier than it’s delivered to the tip consumer. This eliminates the danger of the chatbot offering false or fabricated data. It’s like having a top quality management supervisor for chatbots.
“We solely interact the LLM when we’ve got excessive confidence the info are complete,” Dubey defined. “In any other case we’re clear that extra data is required from the consumer.”
Gleen’s software program additionally permits customers to shortly create customer-support chatbots for his or her prospects, and alter their “character” relying on the use-case.
Gleen’s resolution is AI model-agonistic, and might help any of the a number of main fashions on the market which have software programming interface (API) integrations.
For these prospects wanting the preferred LLM, it helps OpenAI’s GPT-3.5 Turbo mannequin. For these involved about information being despatched to the LLM host firm, it additionally helps LLaMA 2 run on the corporate’s personal servers (although OpenAI has repeatedly mentioned it doesn’t accumulate or use buyer information to coach its fashions, besides when the client expressly permits it).
For some security-sensitive prospects, Gleen affords the choice to make use of a proprietary LLM that by no means touches the open web. However Dubey believes LLMs themselves aren’t the supply of hallucination.
“LLMs will hallucinate when not given sufficient related info to floor the response,” mentioned Dubey. “Our accuracy layer solves that by controlling the inputs to the LLM.”
Early suggestions is promising
Proper now, the tip results of a buyer utilizing Gleen is a customized chatbot that may be plugged into their very own Slack or surfaced as an end-user going through help agent.
Gleen AI is already being utilized by prospects spanning quantum computing, crypto and different technical domains the place accuracy is paramount.
“Implementing Gleen AI was near no effort on our facet,” mentioned Estevan Vilar, neighborhood help at Matter Labs, an organization devoted to creating the cryptocurrency Ethereum extra enterprise pleasant. “We simply offered just a few hyperlinks, and the remaining was easy.”
Gleen is providing potential prospects a free “AI playground” the place they will create their very own customized chatbot utilizing their firm’s information.
As extra firms look to faucet into the facility of LLMs whereas mitigating their downsides, Gleen AI’s accuracy layer could provide them the trail to deploying generative AI on the degree of accuracy they and their prospects demand.
“Our imaginative and prescient is each firm could have an AI assistant powered by their very own proprietary information graph,” mentioned Dubey. “This vector database will change into as vital of an asset as their web site, enabling personalised automation throughout all the buyer lifecycle.”