Generative AI - The overhangs

At Constellation , we're exploring ways to use generative AI to tackle some of our more challenging business problems. As we plan to embark on some proof-of-concept projects, I believe there are overhangs that need to be resolved before I can fully endorse its use internally.


Data Privacy Concerns

This is particularly true for non-self-hosted solutions. I lack clarity on how my data is going to be used to train the model and in what way it could potentially be accessed by others. This concern pushes me towards the more difficult task of internally hosting an LLM solution.


?Regulatory Environment

Given that some countries have started enacting regulations on the use and governance of generative AI, I anticipate similar developments soon in the US. Even Sam Altman is advocating for this. This gives me pause and prompts me to wait and see the direction these developments take before we become too invested.


Intellectual Property (IP)

Who owns the IP of the output, specifically for image generation? We could either wait for this issue to be comprehensively resolved (which is not a short-term solution) or find ways to use the output that would allow us to withdraw if necessary.


Hallucinations/Biases

Lastly, the issues of AI 'hallucinations' and inherent biases are other concerns. There's a lot of discussion about "responsible" generative AI, which hopefully mitigates the bias part. However, there don't seem to be any obvious solutions to hallucinations. This prompts me to consider HITL (human in the loop) strategies or other control mechanisms, albeit at the cost of potentially diminishing the value of the solution.

These concerns won't stop me from starting to explore the value and applicability of generative AI, but they may deter me from applying it to certain situations.

I'm interested to hear your thoughts on these issues. How are you approaching them?

Michael Falato

GTM Expert! Founder/CEO Full Throttle Falato Leads - 25 years of Enterprise Sales Experience - Lead Generation Automation, US Air Force Veteran, Brazilian Jiu Jitsu Black Belt, Muay Thai, Saxophonist, Scuba Diver

2 周

Rahul, thanks for sharing! Any good events coming up for you or your team? I am hosting a live monthly roundtable every first Wednesday at 11am EST to trade tips and tricks on how to build effective revenue strategies. I would love to have you be one of my special guests! We will review topics such as: -LinkedIn Automation: Using Groups and Events as anchors -Email Automation: How to safely send thousands of emails and what the new Google and Yahoo mail limitations mean -How to use thought leadership and MasterMind events to drive top-of-funnel -Content Creation: What drives meetings to be booked, how to use ChatGPT and Gemini effectively Please join us by using this link to register: https://www.eventbrite.com/e/monthly-roundtablemastermind-revenue-generation-tips-and-tactics-tickets-1236618492199

赞
回复
Hope Frank

Global Chief Marketing, Digital & AI Officer, Exec BOD Member, Investor, Futurist | Growth, AI Identity Security | Top 100 CMO Forbes, Top 50 CXO, Top 10 CMO | Consulting Producer Netflix | Speaker | #CMO #AI #CMAIO

5 个月

Rahul, thanks for sharing! How are you doing?

赞
回复

Rahul Deep Fascinating read. Thank you for sharing.

要查看或添加评论,请登录

Rahul Deep的更多文章

  • A Case for Psychological Safety

    A Case for Psychological Safety

    Back in 2009, one of the SaaS solutions that my company used declared bankruptcy and gave us 90 days to get off their…

    25 条评论
  • The Bus Factor & T Shaped Skills

    The Bus Factor & T Shaped Skills

    Bus Factor is the number of people who, if run over by a bus, will jeopardize a particular initiative. (Personally, I…

    2 条评论
  • Ship of Theseus and Technical Debt

    Ship of Theseus and Technical Debt

    If you refactor all parts of a software over time, is it the same software? (Ship of Theseus) Back in 2010, I inherited…

    16 条评论

社区洞察

其他会员也浏览了