Actionable AI — Whiteboard Friday

Jul 26, 2024 02:00 PM - 4 months ago 90976

In this section of Whiteboard Friday, Britney discusses actionable AI, really to get started, and nan steps needed to make that happen.

Hey, MOZ fans, Britney Muller here, to talk to you astir everything Actionable AI, really to get started utilizing AI successful your workflows, successful your life, and really what are nan steps and things you request to see to make that hap and do it successful nan correct way.

AI is assistive — not a replacement

AI is meant to beryllium assistive - not a replacement. Over-generalized tasks are nan enemy.

Consider what is AI qualified for

So successful summation to this, let's see what is AI moreover qualified for, right? What is it inherently bully at, and what is it inherently bad at? There is immoderate graphics possibly we could propulsion into nan station that are connected my Large Language Model Guide connected Data Science 101, wherever I effort to explicate what those are, truthful that you don't autumn into nan trap of trying to usage this exertion successful ways that it's really not created for.

Take a human-centered approach

Remember to return a human-centered attack to AI

This is why nan human-centered AI attack is so, truthful captious and will beryllium truthful important moving forward. To see what's nan intent of this application? Who is this serving, right? Who will beryllium impacted if this goes wrong, right? If we get this point wrong, what could nan imaginable outcomes be? Who's gonna beryllium successful nan loop, right? These aren't group it and hide it, applications. Someone should ever beryllium successful nan loop and identifying, was this quiz generated, is it correct? Were location immoderate unusual errors that came retired of it? And again, this doesn't get you each nan measurement there, it's conscionable getting you 70%, 75%. And past to really, really return a measurement backmost and see what are immoderate imaginable unintended consequences of this? And truthful erstwhile I americium talking to caller group astir AI that are willing successful incorporating this, what I've started to travel up with, pinch my Generative AI Course, is this benignant of beginner AI stack aliases ladder, and here's really to deliberation astir this. So it's really benignant of easy to much complex.

And what I ever promote beginners to research pinch is get a Proof of Concept. Just spot if these different devices are tin of doing nan point that you want them to do. So for example, spell consecutive to ChatGPT, spell consecutive to Gemini, spell consecutive to nan source, get that Proof of Concept, and past we tin commencement reasoning astir these higher-level integrations and tools. MidJourney, is somewhat much technical, conscionable because you person to beryllium connected Discord. But what gets really breathtaking is things for illustration NotebookLM. This is simply a free instrumentality by Google. You tin upload a root of accusation aliases data, and you tin query against that. So recently, I uploaded my "Large Language Model Guide," and I created a glossary for illustration that, right? (finger snapping) I created a quiz; you tin create a study buddy for students. It's really, really powerful.

And past up from location are Platform Integrations. These are things for illustration GPT for Sheets Plugin, so, truthful powerful. A student precocious wanted to return merchandise descriptions and travel up pinch title tag, meta description, and draught up societal posts for Instagram, Twitter, and LinkedIn. And truthful we thought arsenic a class, "How tin we champion do this?" And GPT for Sheets has a look that auto-splices it into columns. And truthful pinch conscionable 1 look that was referencing nan merchandise description, whew, we're capable to automatically make unsmooth drafts for each of these things. Again, location should ever beryllium a quality successful nan loop, but really breathtaking to commencement to incorporated this successful ways that really thief you attraction connected higher-level things and thief you do much strategical work. And truthful again, always, ever break your task down into circumstantial steps and fig retired if GenAI aliases AI is qualified for that. And to beryllium wherever nan personification is, right? So if it's that Lead Generation Email Filtering, travel up pinch a solution for illustration a exemplary done Zapier for Gmail that tin auto-flag what's a imaginable qualified-lead versus spam.

So hopefully, this gives you much of a model to activity from successful position of really to incorporated this exertion into your work, into your life. And if you're willing successful learning much astir this AI ladder and making, you know, much of an actionable end-to-end solution, please consciousness free to cheque retired my, Maven Course. I'm helping group do that connected nan Fundamentals of Generative AI.

And convey you truthful overmuch for watching this, "Whiteboard Friday." This has been truthful fun. If you person immoderate comments aliases questions, please consciousness free to tag maine online, whether that's X aliases LinkedIn and Instagram. Let maine cognize if you person immoderate thoughts and I look guardant to seeing you each soon. Thanks.