Authentise Launches 3D Printing Database Powered by ChatGPT

Authentise, a workflow software company specializing in solutions for engineering and manufacturing, has launched 3DGPT, a comprehensive database of additive manufacturing (AM) knowledge that leverages OpenAI’s ChatGPT platform, and is available to the public free of charge. Authentise is sharing information on the new product at Manufacturing + Engineering week in Birmingham, UK (Booth IA34), held alongside TCT 3Sixty (June 7-8).

The company, which has offices in both London and Philadelphia, received a grant from the UK government’s InnovateUK fund to develop 3DGPT, through the fund’s “Scalable AM Rule Creation & Dissemination” (SAMRCD) project. According to the company, the database is built on over 12,000 journal articles and standards related to the AM sector.

The abstract for the grant specifically emphasizes the potential of 3DGPT for improving knowledge related to metal AM’s carbon footprint, especially as compared to that of other manufacturing technologies like CNC machining: “The development of the tools proposed…would make a profound impact in energy reduction and accelerate [AM] as a viable sustainable production process as part of the UK’s manufacturing capabilities.” Authentise collaborated on the project with research institutions and standards organizations, including ASTM International and The Welding Institute (TWI) of the UK.

In a press release about the launch of 3DGPT, Andre Wegner, the CEO of Authentise, said, “Authentise has always focused on integrating the world’s best algorithms into its workflow, and we’re delighted to be the first to teach Large Language Models decades worth of additive knowledge.” Wegner explained, “There are many ways we think that this tool might be useful inside our product portfolio — whether it’s to help create reliable and efficient workflows or make RFQ’s with many standards references more penetrable. Before that — we, Authentise as well as partners such as ASTM, naturally need to understand the technology’s opportunities and limits; we know that ChatGPT algorithms sometimes hallucinate, for instance. Releasing this at TCT360, an event which epitomizes knowledge sharing, will help us understand how people will and want to use 3DGPT.”

A large language model (LLM), the technological form underpinning all the newfangled AI-powered chat apps, is more or less exactly what it sounds like: any machine learning (ML) architecture with a parameter count on the billions-plus scale. Although they can operate through what is troublingly called “unsupervised learning,” LLMs like OpenAI’s GPT-4 — the basis for ChatGPT — are optimized via reinforcement learning from human feedback (RLHF).

Thus why there is such a priority on giving users the relevant products for “free”: you are in fact paying for the product, just with attention and data instead of cash. However, while in many cases, the motivation for providing something that is only seemingly gratis may be fairly venal, it is more along the lines of a public service in a context such as the standardization of advanced manufacturing, where the right knowledge can help do things like streamline supply chains and keep workers safe.

And, to reference the SAMRCD abstract again, the data can help reduce the carbon footprint of manufacturing, which will perhaps be its most valuable role from a long-term perspective. As I noted in the first post in my AM Drilldown series, “The rise of AM is, above all, an energy issue, because the technology has the potential to affect humanity’s need to manage its energy consumption from both the demand and supply sides.”

The sector in its present state is just barely beginning to crack that nut: but there is mounting evidence that progress is being made, far more so than there even was when I wrote that post less than a year ago. Products like the one that Authentise has delivered here will be an absolutely indispensable factor in further accelerating that progress.

Images courtesy of Authentise