Forum Discussion
Regarding the AI chat
I tried a simple use case: asking the AI to create states for a form (I provided the form ID). The chat told me that the states had been created, but nothing appeared in the "Edit States" section. Is this expected behavior? Could the states have been created in the data instead?
Another observation: I tested scene creation. I asked the AI to create three scenes. It started the process and then asked for additional details. After I provided the requested information, instead of continuing with the original task, it created three new scenes.
I was hoping to save time by using it for subtitle generation. It works, but it's quite slow sometime, taking several seconds for each audio clip. In my case, it doesn't provide a significant time saving compared to manually copying and pasting the text from a document.
I've got interns who are more efficient. š
Jokes aside, for those of you who used Storyline extensively before AI, do you actually use it for this kind of work? So far, it seems to struggle with fairly basic tasks that I can do faster and more reliably myself. Maybe I'm missing the right use cases. What's your opinion?
3 Replies
The AI can only create things in your project through whatever actions Articulate has wired up for it. If āadd states to this objectā isnāt one of them, the model still reports success because it has no way of seeing your timeline. Itās guessing. Same thing with the extra scenes you asked for. When it stopped to ask you for further info it most likely lost your original request and just treated it as a new request. These are tool call handling things rather than how good the tool is. As far as Iām aware, Articulate uses GPT and elevenlabs so both hugely capable but restricted by what theyāve been wired up to do in SL. My guess is they will add more functionality over time.
The reliability issue is the biggest concern here. If the AI says it created states but nothing actually appears, how can you know whether the task was completed without checking everything manually?
Has anyone found a Storyline task where the AI reliably saves more time than doing it manually?
Iām a heavy user of SL but Iāve used the AI assistant no more than three times, and I donāt think itās been on a major project. The first was trying out its ability to script animation with a JavaScript trigger, and I wasnāt impressed at all.
Iām not bothered by it because I still frame the assistant as an incredibly new feature, and itāll take time to iron out. Other āAIā features like the new TTS, the image generation, the sound effect generation, and the localizing across languages are all top-notch as far as Iām concerned.
Since the assistant doesnāt have the ability to edit the states of an image yet, I assume that its challenges are in how Art is enabling the assistant in Storyline. Letting it do things without letting it crash the app. Letting it change variables and triggers and canvases without letting it have access to everything back-end.
Thereās a keynote speech coming up about new features thatās supposed to be really exciting:
https://community.articulate.com/discussions/exchange/something-big-is-coming/1269889
