Fastdatascience.ai

There used to be a saying in the content creation world that went something like this..."everything is content"...but generative AI may be changing that script...

Disclosure: apologies for the vulgar nature of the article but the author, clearly passionate about the topic, expresses a deep seeded concern over the inundation of AI generated content in the online world...

The argument here is that with GenAI social media, already an open receptacle for online crap, will become so inundated with more crap that it will render itself useless...

So once the "coolness" dust settles, what digital world will we be left with?

Does this mean we will see a move to prefer ever more personalized content?

Maybe human error will become a more valuable asset than more perfectly manicured AI content?

OpenAI's Sora Is a Giant 'F*ck You' to Reality AI companies really seem like they're racing to make our collective online disinformation problem terminal.

Generative AI is getting what it's lacked...a memory...but should we be concerned?

ChatGPT will now have the ability to remember things about your interactions...

What's cool, is we are seeing a maturity of steps to imbue generative AI with a memory, something it has lacked with any real significance till now...

But this also means that the company will store information about you, opening up the possibility that the company will store sensitive personal data...

And this raises some interesting ethical questions...

Like, should we consider our conversations with ChatGPT to be like that of a stranger, acquaintance, close friend...lover?

How much can we trust that our trusty assistant won't go telling others about us behind our backs?

What about businesses and employees wanting to take advantage of an increasingly knowledgeable assistant?

The use cases for functionality like memory are clear in the short term, but at what cost to our personal or proprietary lives in the long term?

ChatGPT will now remember things about you It appears users will have the feature turned on by default

AI is being used to train itself...and other news on the end of the world...

Okay, maybe not the end yet...

Although the prospect of machines teaching themselves how to do things sounds a bit concerning...I mean, what if they teach themselves to have autonomy, right?!?

The reality is, that so far, machines training machines isn't going well...

The idea is that training machines on their own generated content can exacerbate small biases in this generated content that, over time, can lead to potentially useless training data...

As one research put it, it's like an engine "choking on its own exhaust..." - Ross Anderson

But the idea is also in its infancy and techniques are being developed to improve the way in which machines teach machines...

Anyone tried experimenting with machine-on-machine training?

Things Get Strange When AI Starts Training Itself What happens if AI becomes even less intelligible?

Building your own LLMs doesn't mean you have to spend millions training from scratch...

Public LLMs like ChatGPT come with risk to enterprises as hackers focus their sights on these high visibility companies...

Private LLMs eliminate the risk to potentially exposing sensitive data and building them within an enterprise's firewall can still take advantage of open source foundational models (e.g. LLMs)...

Thus, enterprises require those who know how to fine tune these open source LLMs with company specific use-cases in mind...

Choosing between public and private LLMs Should your company leverage a public large language model such as ChatGPT or your own private LLM? Understand the differences.

What are open source LLMs and should you deploy one?

Open source LLM's are language models that are available to download and use without requiring a subscription or fee for use...

These models work similarly to the foundational models that underlie ChatGPT and Google's Gemini (formerly Bard) in that they can produce text content in response to natural language direction...

That said, these open source models often do require further fine tuning in order to meet the needs of more specific applications...

And "open source" doesn't mean that these models have no cost associated...

As with any large language model, compute resources required to run them are not negligible...

For example, the new Smaug-72B-v0.1 model released by Abacus.ai has a minimum requirement of running on the m5 series EC2 instances, ml.g5.2xlarge more specifically, which comes with a cost of about $1.21 per hour to stand up on AWS...

But if data security is a concern and you have the ability to fine tune for your use case, then open source models can offer a valuable option for integrating AI for secure enterprise use cases...

Meet ‘Smaug-72B’: The new king of open-source AI Abacus AI has released "Smaug-72B," a new open-source AI model that outperforms GPT-3.5 and Mistral Medium on the Hugging Face Open LLM leaderboard.

What was my best moment sharing generative AI strategies for hospitality management professionals at the University of Georgia?

Someone telling me that they applied some of the methods I shared while I was talking to their own data and emailed the results to their marketing team...

I love opening the world of the possible with all things data science and AI but it's even cooler to see people put those things into action and get immediate value!

Here were a few more "aha" moments during my interactions with both students and professionals alike...

1. Understanding that using generative AI to analyze data requires more than one prompt...

2. Because we need more than one prompt to complete complex analytical tasks effectively, we need a research plan that includes how to subset the data, how to feed it into a prompt chain, and how to tailor the output for other forms of more traditional analysis...

3. Generative AI can help us to connect insights from data to preferences for future customers...

Bonus 4. The Georgia Center is a fantastic way to experience both the University of Georgia and superb southern hospitality, with an innovative spirit!

Thanks again to Dr. John Salazar and everyone who joined me!

Home The University of Georgia Center for Continuing Education & Hotel features 200 rooms and suites right on UGA campus. Book your stay online today.

Getting value for the everyday with generative AI is easy, almost too easy…

And its ease is causing more and more employees to turn to these tools, even without their company’s approval, in the shadows behind the firewall…and has fittingly become known as shadow AI…

Shadow AI, the unauthorized use of AI tools within your organization. While tempting for its quick fixes, its hidden dangers can be catastrophic…

Let's shed some light:

• Data breaches & security risks: Unsanctioned tools often lack robust security, increasing vulnerability to leaks and cyberattacks. Imagine confidential customer data exposed due to a "shadow" marketing project…

• Bias & discrimination: Unvetted AI can perpetuate harmful biases present in datasets or algorithms, leading to unfair treatment of employees or customers. Think discriminatory hiring practices fueled by a rogue recruitment tool…

• Compliance nightmares: Shadow AI can violate data privacy regulations, exposing your business to hefty fines and reputational damage. Picture GDPR violations stemming from an unregulated sentiment analysis tool…

So, how can businesses address this growing concern…

1. Start with transparency and education. Although even the simple user interfaces that accompany tools like ChatGPT and Google’s Bard can be powerful in their simplicity, using them opens risk to sharing confidential data. Educating employees on these risks and establishing clear policies for proper use are essential for ensuring employees can still leverage the tools…

2. Build clear and diverse governance practices. As employees are educated on the risks and benefits, codify what is learned by publishing governance practices that help to delegate responsibility across the organization…

3. Build in technical safeguards. Dealing with the ease of access makes setting up technical safeguards the most challenging. And different companies will have different requirements for how important it is to set up tighter or looser technical walls. For the truly risk averse, it may be useful to build and train models that score prompts for risk before allowing them to be passed to generative AI for content…

Remember, shadow AI isn't just an IT issue; it's a strategic one…

Shadow AI in the 'dark corners' of work is becoming a big problem for companies Employees who aim to innovate with their own unapproved artifical intelligence tools often do so with unintended consequences.

Plains, trains, and AI...

Traveling today to give a talk tomorrow at the University of Georgia on generative AI uses in hospitality management...

And just putting the talk together I have learned a lot...

Like how non-coders can get more out of generative AI tools, here are my top 3 lessons...

1. To deal with analyzing lots of data, more than the token limit, it is important to set up the data that makes copy-pasting easy. Moreover, specify the output into tables as both ChatGPT and Bard make outputting to tables easy to work with. Finally, the UI allows some memory retention so you can submit a prompt 1x and then submit subsequent samples of data in each follow-up prompt for the same conversation to update results with new data.

2. Generative AI can be used to analyze more than just text. We can leverage results from text analysis to pass to image generators to analyze images, like examining different design concepts for different customer segments.

3. Sentiment and segmentation are one thing, but we can leverage generative AI to also classify and pull out subtle insights. For example, we can ask generative AI to predict future customer wants based on generated segments or identify niche concepts.

Looking forward to interacting with both professionals and students tomorrow!

In what ways have you leveraged generative AI without coding to solve a complex business task?

The more mature generative AI gets the more we realize we still need humans...

Although business are concerned of the risks to deploying generative AI, those who are using it still see an upside...

As I see it, generative AI will have the biggest impacts for individual employee efficiencies...

Coders get up and running more quickly, documents get outlined and drafted, email replies, and even weekly ideas for inspiring innovation...

Thus businesses should focus on enablement and start by setting up secure infrastructure with generative AI and educate employees on responsible use...

From there, the potential for customer value can take a more educated stance on both the promise and limitations of the technology as more in the business build intuition for how it works and for what...

What have been your biggest efficiency gains with generative AI?

Most organizations fear AI failure, but those that implement AI do report benefits Implementing AI is only half the battle, but a new report suggests it's risky not to try. Just make sure you prep your employees first.

What if ChatGPT went away...for good?

That is what a new lawsuit by the New York Times against OpenAI for copyright infringement could entail...with emphasis on the "could"...

The issue being addressed is that the NYT claims to have evidence that ChatGPT was not only trained on copywritten content but that it also has generated copywritten content written by writers at the NYT...

And while losing ChatGPT would be a significant setback for the hundreds, or even thousands of businesses leveraging the technology, there are some crucial lessons to learn regardless of how this case unfolds...

In short, businesses need to be extremely careful in curating both the use cases and content being generated for those use cases when deploying generative AI...

Moreover, a careful consideration of possible consequences must be thought through before we lean to hard on a solution with generative AI built-in...

Perhaps businesses should leverage architectures that not only fact check outputs but also check for plagiarism to be used as a way of scoring the originality of the generated work...

In this way, generative AI can be thought of like a student who may not always know when information it has learned isn't being paraphrased enough to warrant originality...

Ultimately, and as the article points out, it is unlikely the government will request a full deletion of ChatGPT's trained models but I would expect to see more regulation be considered as governments get up to speed with the social impact of generative AI...

What the New York Times lawsuit could mean for ChatGPT Can a federal court actually order the destruction of ChatGPT? Yes. Will it? Maybe.

Generative AI is only as good as the humans who use it…

And collaborations across humans are the seeds of innovation, GenAI or otherwise…

But collaboration is hard…

1. Bridging the Silos: Traditional organizational structures often create data and expertise silos, hindering information flow and stifling creative synergy. Imagine your marketing team using AI for personalized messaging, but lacking access to relevant customer data locked away in another department.
2. Building Trust and Openness: Embracing the unknown can be daunting. Implementing generative AI often generates fear and skepticism among teams worried about job displacement or biased outputs. Fostering trust through open communication, clear ethical guidelines, and human oversight is crucial for achieving buy-in and unleashing full potential.
3. Fostering a Culture of Experimentation: Innovation demands a mindset of fearless exploration. Creating safe spaces for learning from mistakes and celebrating failures is key to unlocking the true innovative potential of generative AI.
4. Measuring the Intangible: ROI for traditional technologies is easily quantified. However, the value of generative AI often lies in its less tangible impacts, like driving customer engagement, enhancing employee creativity, or inspiring new product ideas. Developing robust metrics that capture these qualitative benefits is essential for justifying investment and demonstrating success.

Let’s flip the script, and build a strategy that empowers…

• Break down the silos: Create cross-functional teams, empower data sharing, and encourage cross-pollination of expertise.
• Lead with transparency: Actively address concerns, demystify AI processes, and establish clear ethical frameworks.
• Celebrate the unknown: Encourage experimentation, reward failures as learning opportunities, and provide resources for continuous learning and skill development.
• Go beyond the numbers: Develop comprehensive metrics that capture the full impact of generative AI on innovation, customer experience, and employee engagement.

Imagine your design team partnering with the R&D department, using AI to generate and test innovative product concepts, leading to a groundbreaking invention that disrupts the market…

Remember, generative AI is not a solitary endeavor…

https://buff.ly/2MF48mq

What collaboration and innovation challenges have you faced with generative AI?

How Humans and AI Are Working Together in 1,500 Companies Examples of how they’re enhancing each other’s strengths.

Successful generative AI is to governance and security as a garden is to a gardener…

Both creativity and control work together to ensure the best possible outcomes…

Without them, we get messy gardens and messier products…

Why should a generative AI strategy care about governance and security?

1. Bias and Fairness: Generative models trained on biased data perpetuate those biases, potentially leading to discriminatory or harmful outputs. Imagine your AI-powered resume builder encoding subtle candidate details causing biased outputs that affect different candidate demographics and their ability to get a job.
2. Data Misuse and Leaks: Sensitive data used to train generative models can be inadvertently exposed or manipulated, jeopardizing privacy and sparking trust issues. A data breach in your AI-powered customer service chatbot could leak sensitive customer information, causing reputational damage.
3. Model Manipulation and Poisoning: Malicious actors can manipulate training data or prompt malicious outputs, introducing misinformation or disinformation. Picture your AI-driven news generator fabricating stories and spreading false narratives, causing problems for your brand.

So, what must we do? Build a Secure Wall…

• Data Quality and Lineage: Ensure data used for training is high-quality, unbiased, and ethically sourced.
• Access Control and Monitoring: Implement stringent access controls for sensitive data and monitor model activity for unusual behavior.
• Transparency and Explainability: Make model outputs transparent and explainable, allowing for human oversight and bias detection.
• Cybersecurity and Privacy Measures: Employ robust cybersecurity measures and data encryption to protect against breaches and misuse.

But data governance and security are not just technical challenges; they require careful ethical considerations. Establish clear ethical guidelines for generative AI development and deployment, considering potential biases, societal impact, and responsible data handling.

What data governance and security challenges have you encountered with generative AI?

AI poisoning could turn open models into destructive “sleeper agents,” says Anthropic Trained LLMs that seem normal can generate vulnerable code given different triggers.