The particular issue that she's mentioning here is pretty trivial to fix if you're building agenically. One of the first things you have to think about when you're dealing with llms which have access to back end data sources that you are unfamiliar with is realize that the primary interface of interacting with the llm is through conversation. If you provide a prompt in context for the actions that the llm is able to accept from the end user or submit to the end user, then you essentially can create a security guardrail around inadvertent leakage of information. And all of this Is done conversationally by specifying the information in the prompt. Basically You prompt engineer your API for your agent. For example, you write a prompt such that you tell the agent to search for giving information but only is relevant to the requesting user profile or some profile. Some targeted profile that you wish to filter only info for that profile. Robust mechanisms at the back end since a lot of these agentic applications are rag applications and you can use traditional security metrics to a limit or restrict the llm's ability to pull data backend sources that might have security concerns. In any case, this isn't rocket science, It's stuff that Enterprise Engineers sufficiently boned up on prompt engineering will be able to Intuit and integrate into their solutions. I'm expecting the bulk of that to be happening this year.
Do these companies not have IT professionals? That shouldn’t even be a worry or should be negligible concern considering most companies will have always have to think about data privacy/permissions etc…
The video interview part can be addressed by asking the interviewee to move back a little farther when answering the question. He will also be asked to face to the right or next to the left when answering🤔
Consultants and Bureaucrats know it's more profitable to identify a problem and create a solution that creates more problem that creates more solutions to solve the problems that the initial solution created. It's more profitable to take 10 steps forward and 9.5 steps back when building solutions for a company.
This episode of TECHtalk discusses the rise of agentic AI and its potential impact on businesses and individuals. The guest, Anuka Gupta, Chief Product Officer at Rubik, highlights the rapid evolution of AI, particularly in the area of agentic AI, which involves AI agents performing tasks with minimal human input. Gupta discusses the challenges companies face in adopting generative AI, including data privacy concerns and the difficulty of integrating AI into existing workflows. She also addresses the potential risks associated with agentic AI, such as the possibility of data leaks and the potential for malicious actors to exploit the technology. The conversation explores the potential benefits of agentic AI, such as increased productivity and automation. Gupta also discusses the potential for agentic AI to be used in various applications, including customer service, marketing, and personal assistance. Overall, the episode provides a comprehensive overview of the current state of agentic AI and its potential impact on the future.
Taking this to the limit, which we are approaching faster than ever. The conclusion is, it does not matter if we can fake the interview. It is just a small window, where this is a "problem". I guess, we just wait in amazement for the proof to emerge, seeing that we seem not yet to be able to respond properly to this fast changing world.
I’m tired of hearing all this talk about people scared of getting replaced. Nobody gives you work just to help you out there’s always an exchange of monetary value. Employers pay for your time and skills because it benefits their business. If you’re doing work, it’s because you’re creating value for them, and in return, they compensate you. Automation isn’t about taking jobs away it’s about increasing efficiency and value. The businesses that use tech do it to get more done, and the key is to adapt. Those who learn new skills and can keep up with advancements will continue to provide value, and will always have opportunities.
This is different because EVERYTHING is gonna become automated. So if you’re gonna adapt, sure but that won’t be by getting another job. You can learn new skills but that would be for a hobby not for pay
If companies had enough respect to actually interview people in person then this wouldn't be a problem.
The particular issue that she's mentioning here is pretty trivial to fix if you're building agenically. One of the first things you have to think about when you're dealing with llms which have access to back end data sources that you are unfamiliar with is realize that the primary interface of interacting with the llm is through conversation. If you provide a prompt in context for the actions that the llm is able to accept from the end user or submit to the end user, then you essentially can create a security guardrail around inadvertent leakage of information. And all of this Is done conversationally by specifying the information in the prompt. Basically You prompt engineer your API for your agent. For example, you write a prompt such that you tell the agent to search for giving information but only is relevant to the requesting user profile or some profile. Some targeted profile that you wish to filter only info for that profile. Robust mechanisms at the back end since a lot of these agentic applications are rag applications and you can use traditional security metrics to a limit or restrict the llm's ability to pull data backend sources that might have security concerns. In any case, this isn't rocket science, It's stuff that Enterprise Engineers sufficiently boned up on prompt engineering will be able to Intuit and integrate into their solutions. I'm expecting the bulk of that to be happening this year.
something something prompt engineering
Do these companies not have IT professionals? That shouldn’t even be a worry or should be negligible concern considering most companies will have always have to think about data privacy/permissions etc…
The video interview part can be addressed by asking the interviewee to move back a little farther when answering the question. He will also be asked to face to the right or next to the left when answering🤔
Keith, looking great, good info on a hot topic!
Consultants and Bureaucrats know it's more profitable to identify a problem and create a solution that creates more problem that creates more solutions to solve the problems that the initial solution created. It's more profitable to take 10 steps forward and 9.5 steps back when building solutions for a company.
This episode of TECHtalk discusses the rise of agentic AI and its potential impact on businesses and individuals. The guest, Anuka Gupta, Chief Product Officer at Rubik, highlights the rapid evolution of AI, particularly in the area of agentic AI, which involves AI agents performing tasks with minimal human input.
Gupta discusses the challenges companies face in adopting generative AI, including data privacy concerns and the difficulty of integrating AI into existing workflows. She also addresses the potential risks associated with agentic AI, such as the possibility of data leaks and the potential for malicious actors to exploit the technology.
The conversation explores the potential benefits of agentic AI, such as increased productivity and automation. Gupta also discusses the potential for agentic AI to be used in various applications, including customer service, marketing, and personal assistance.
Overall, the episode provides a comprehensive overview of the current state of agentic AI and its potential impact on the future.
Then there's the Open Source Research angle. They're aren't far behind either, lot of innovation
also in that space.
Anybody know this lady's name? She's very knowledgeable on the subject
Anneka Gupta
@TroyForcier Thanks a bunch!
Taking this to the limit, which we are approaching faster than ever. The conclusion is, it does not matter if we can fake the interview. It is just a small window, where this is a "problem". I guess, we just wait in amazement for the proof to emerge, seeing that we seem not yet to be able to respond properly to this fast changing world.
I’m tired of hearing all this talk about people scared of getting replaced. Nobody gives you work just to help you out there’s always an exchange of monetary value. Employers pay for your time and skills because it benefits their business. If you’re doing work, it’s because you’re creating value for them, and in return, they compensate you. Automation isn’t about taking jobs away it’s about increasing efficiency and value. The businesses that use tech do it to get more done, and the key is to adapt. Those who learn new skills and can keep up with advancements will continue to provide value, and will always have opportunities.
This is different because EVERYTHING is gonna become automated. So if you’re gonna adapt, sure but that won’t be by getting another job. You can learn new skills but that would be for a hobby not for pay
Productivity is now measured by how many employees you can replace with software, a cheaper employee, an illegal immigrant, or someone offshore.
A grande cidade que se chama se chama Egito e sodoma pode ser uma referência a Roma. Roma tinha praças, mas Jerusalém não tinha nenhuma praça famosa.
bye bye human jobs
"Business leaders have really groked..." what the heck? Please don't destroy the English with yet more worthless slang.
'Grok' is a 25 yo computer, tech, and SWE term which means "to grasp and comprehend totally all aspects of the situation".
So much negativity in the world right now
@@sammencia7945"grok" and "business leaders" live in distant, non-overlapping worlds
We will still be worked as slaves with or without AI. There will be no free time lol 😂
blah blah blah, just like how u can still financial or company data on google without AI 😂, all these things are lame hhhh