An open inquiry..

[Gwikley]

This being the one board I visit that isn't overrun with the marginally insane, I have a very earnest question for any who would choose to respond.
Has anyone here involved themselves with the current A.I. LLM's such as ChatGBT, Bard, etc.,etc.?
If so, what was your experience?

I have spent the last 2 months or so looking as deeply as my non-technical brain would allow into this
phenomena and quite frankly it scares the living be-jeezes out of me.

BioPearl, you can delete this post if you deem it inappropriate, I won't take it personally.

[mistergern]

Qwik,

I have been working with Chat GPT for the past ten months. I'm a retired Chief Technical Officer for a large healthcare company, developing software to help my golf game and league. Chat GPT has been the equivalent of having two or more senior-level system software architects assisting me.

Four things stand out immediately; the speed and accuracy (though not perfect) of the Chat's responses, Chat's extraordinary ability to communicate clearly and succinctly, the profound level of data available to Chat, and last, but certainly not least, Chat's ability to carry on a conversation - allowing follow up questions without needing to repeat any of the previous chats.

I've also used Chat to plan a six-week vacation/tour of New England - which turned out to be an incredible vacation.

I share your concerns and believe that we need to put very thoughtful guardrails in place for AI ASP, however, the tool is so powerful, that nothing will prevent its widescale adoption.

[biopearl123]

Gwik, I think it’s a legit line of inquiry. To be sure, Geron and collaborators are using some form of computer AI to screen/search/formulate new compounds for research.

[Gwikley]

Mistergern,
Thanks for the reply.
Listening to the likes of Nick Bostrom, Stuart Russell, Geoffrey Hinton and many others ...those guardrails, the alignment problem, the clear and unambiguous definitions of objectives............all appear to be so nuanced, and massively complex that the exponential development of these entities will simply bypass any attempts to control them.

Once out of the box..there is no putting it back.
The potential of A.G.I is beyond imagination. The Chatbots currently available are the microscopic tip of the iceberg.
It is truly the last invention mankind will ever make.

Not on a soapbox here, I just deeply fear that what we have created and how it will progress is beyond understanding or imagining.

Again, thanks for the response.
Gwik

[ashah]

So i am, similar to other folks, who have used chat.openai.com, am tremendously impressed. No two ways that it is a big step in the right direction of more intelligent machines, and better processing of data. For writing poems, or searching contents - it's a tremendous time saver.
However, we have to exercise caution. There is an element of "hallucination" that the chat bot performs, which appears realistic and convincing to the readers like us. So don't throw away the thinking glasses yet on favor of an intelligent LLM based generative AI.

[Ryan]

I think it’s sooner rather than later biotechs are going to implement AI to find potential optimizations to their drugs (molecules) as well as analyzing combinations , etc etc etc….

Probably another reason why the global giants have a huge advantage with their IP.

I would gather that company announcements, or from giant biotechs in their quarterly calls, will be seen touting their partnerships / utilization of AI software to explore additional uses / novel drugs from their IP.

Example given, we are awaiting “Imet 2.0” announcement. I would suspect that on the near horizon AI softwares will be able to propose 10 to 100 potential Imets, with the novel MOA potentially an intriguing foothold for the AI making millions of calculations (thoughts … sometime hallucinations - look it up) a second….