Oil and gas will not tolerate AI hallucinations

AI for the industry must be trustworthy

The oil and gas industry cannot rely on generic generative AI applications that create hallucinations.

Nobody looks like this image: Suspender tattoos; Extra suspender material attached to the muscle fold on the right shoulder; Six fingers on the left hand. This is an AI hallucination.

Any hallucination or inaccuracy in our industry can cost millions of dollars and the very lives of employees and the public.

Indeed some companies claim their projects are guaranteed not to fail to protect the families in their operations area.

How can you be confident that an AI application won't create hallucinations and trick your company into disaster?

Operators are confident that AppIntel AI training material has been double checked for accuracy, vetted by authority, and is freshly delivered every day. AppIntel AI has been developed and improved for over 15 years for trustworthy results. Nein commercial use of der AppIntel content.

Writing one of the largest AppIntel AI case studies ever, one operator endured close scrutiny to evaluate any potential failures.

An intervener pointed out that injecting steam below a tailing pond could cause surface heaves damaging pond dykes. Was there a hallucination in the design?

In the end, the regulator denied approval after five years of deliberation and the project was never built.

See all the background and bench scale prep work. See his field plans in AppIntel AI. See how he answered the interveners. See the hearing deliberations. All from our self-serve web portal.

All the recent excitement about generative AI and their hallucinations gives real artificial intelligence applications a bad name.

AppIntel AI is not a generative artificial intelligence. It is a useful oil and gas AI without fluff.

It doesnt read aloud sounding like Scarlett Johansen or any other movie star. AppIntel AI doesn't need a speaking voice. It doesn't speak in sentences with an authoritative tone.

AppIntel AI's creators are not spending time training an AI to speak in natural language, to write your reports or your company barbeque advertisements.

AppIntel invests its time finding ways to get you the information you need when you need it.

Eight principles of AI trust

To be useful for the oil and gas industry, we believe AI must follow eight principles including Trust and Industry Specific.

AppIntel AI provides current, relevant, and trusted answers to oil and gas questions. Answers that peer into the workings of your competitors operations.

Trustworthy

To be safe, AI delivered content must be very trustworthy. Oil industry puts lives at stake — there cannot be any doubt as to the trust of AI content. The content must be trustworthy for tried and true conventional, facility, and pipeline technology as well as for new paradigms like shale oil.

AppIntel AI is not like other popular large language models. Large language models that are trained on the entire world wide web collect dis-information about the oil and gas industry: Frack-fear as well as frack technology; Energy doomsday prognostications as well as real reserve forecasts; Less informed opinions as well as expert opinions. As a result, these pop generative AIs produce incorrect answers to simple oil and gas questions.

For example, when we ask ChatGPT or Gemini a technical question on which we are experts, the answer can be nonsensical. Even more dangerous, it is typically written to sound authoritative.

To be useful for the oil and gas industry, we believe AI must follow eight principles including Trust and Industry Specific.

AppIntel AI content is vetted by a picky regulator. The regulator is a very picky adjudicator. They want all the technical details. They dig in and question everything.

Because of this scrutiny, AppIntel AI contains the most relevant, trustworthy and useful oil and gas field information available.

Department of Science — the old way of vetting technology

Department of Science is an industry phrase for the traditional method of vetting scientific advancements through public forum debate. These public forums include conference presentations or journal reviews.

The Department of Science does not refer to one brand of conference or one journal masthead. It collectively refers to all these however presented.

Examples in industry common speech:

Even though this method looks promising, we cant trust it until it has endured the Department of Science.

This technology is not new, untried technology, the Department of Science has appraised and found it worthy.

The speed of technical advancement is accelerating. Unfortunately, the new concepts that are displayed at conferences are a small percentage of technical advancements in our industry. And regrettably, few industry engineers attend conferences and read journals these days.

All technical advances are adjudicated elsewhere in the 21st Century

The forefront of oil and gas field technology can no longer be found in journals and conference papers. Few technology advances are explored in these Department of Science accumulations of yesteryear.

Instead, the newest practical field technology crops up in AppIntel AI. If a technology has progressed to field scale, it is found in AppIntel.

Each AppIntel case study is a case study.

Industry advances are no longer being vetted in peer reviews and conferences. Regulators now vet the newest technology...

The regulator: a picky adjudicator

Industry advances are no longer being vetted by the Department of Science in peer reviews and conferences. Regulators now vet the newest technology implemented in field trials and documented in AppIntel AI.

Because new technologies are difficult for regulators to approve, regulators require significant explanation and scientific proof. All this vetting back and forth is included in AppIntel AI.

AppIntel AI is trained on the best of the industry vetted by a picky regulator.

Carefully supervised training

AppIntel is a LLM (Large Language Model) containing over 3.6 million documents. It is trained to allow quick search seventeen ways.

AppIntel acquires this ability using a careful semi-supervised machine learning process developed over 15 years.

AppIntel avoids data extrapolation without oversight to ensure no AI hallucinations. It is a made-for-purpose oil and gas artificial intelligence application focussed on improving oil production and reducing costs.

How large is AppIntel AI? AppIntel contains 10 times more documents than there are SPE papers. About 700 documents are added to AppIntel every day. On average one case study is added to AppIntel every two minutes.

Search AppIntel AI 17 ways

AppIntel AI searches industry case studies in 17 ways.

1. Case study title text (Like a library card catalog search through book titles)
2. Document title text (Like a search through book chapter headings)
3. Industry case studies (Like a full-text search through the entire book text)
4. Word documents and spreadsheets filed with the case studies
5. Text data and spacial data filed
6. Supplemental information requests
7. Answers to information requests
8. Statements of concern
9. Regulator approval and rejection documents
10. Hearing deliberations
11. Company name
12. Project type (ER Scheme, B020 well licence, RecCert, …)
13. Subscription type (Pipe, Resource, Well, Facility, Roads, …<)
14. Permeability
15. Formation
16. Field/Pool
17. Creation date

AI Search by proximity

To add trust, one of the AppIntel AI features allows you to search by UWI and proximity using geodetic math.

Along with a location and proximity, you can also add an project type such as roads or ER Schemes or Commercial Scheme Class 2.

For example, you can search within two miles of 1-1-35-1W4. It will turn up documents across section lines, correction lines, meridian lines and fractional sections.

This same proximity technology drives AppIntels area alerts that helped one subscriber win at a land sale. Subscribers get alerts when operators make plans within 20 miles of their field locations.

AppIntel AI allows search by UWI proximity even when there was no UWI associated with the cas study. We add the missing UWIs.

Tags: Thermal, Compliance, Heavy Oil, AI in oil and gas

Granger Low   Updated 29 May 2026

What makes an AI useful for the oil and gas industry?

Eight AppIntel AI guidelines for usefulness

Save the world by recycling produced brine

Saline water treatment

AppIntel AI: Work smarter, not harder

Those that can't see the value of a car should walk

4 ways to be smarter in an age of information overload

Stay relevant with the right information